In this issue

• Brace for AI’s Big Bang

• New study finds AI boosts work quality by 40%

• Google’s new chatbot ‘5x better’ than ChatGPT

• ChatGPT gets ears, eyes and a voice

• Scientists find telling AI to chill out works wonders

• Why we’ve created The AI Writer

LATEST DEVELOPMENTS

Brace for millions of AI reports, emails and presentations

Ivan Kruk / Shutterstock

We’ve quietly entered the next big phase of the workplace AI revolution. And bots that write your reports and email replies are just the half of it.

It may have been the quietest of Big Bangs, but it was as big as they come. Because we’ve just entered the next stage of the AI revolution.

Last month saw a surge in AI developments that will change how we work forever. And we’re not talking about the distant future either. One of the biggest changes of all is imminent.

ChatGPT’s decline in user numbers now seems to have been just a seasonal blip. Students were on summer break and no longer needed it for their homework. Now, like the rest of us, they’re back at their laptops, with work to do – work they know AI can help with.

Nor is it just ChatGPT itself. Every app from card-payments giant Stripe to Zoom and Duolingo are using the tech behind the chatbot, GPT-4, to enhance their platforms. Little wonder that OpenAI, its parent company, now expects to generate $1 billion in revenue over the next year.

AI writing in weeks

But it was another launch, by Microsoft (OpenAI’s biggest investor), that should have created the biggest stir. After months of speculation, the company has announced that, next month, it will finally switch on Copilot, the AI sidekick for its Microsoft 365 suite of work apps.

That means that, within weeks, more than 200 million people who use apps like Word and PowerPoint will start to get access to a bot that can reply to emails, write reports and create entire presentations for them.

I saw Copilot demonstrated in an internal meeting of leaders at a large oil company in London recently. There were audible gasps around the table when attendees saw it generate whole paragraphs and entire slides from just a few words.

Microsoft has now even added a ‘sound like me’ feature, which will analyse your own writing style and (so the company claims) reply in your written voice.

Will it help?

It’s hard to say how much this will help those of us who rely on the software. Writing is hard, and writing matters. So anything that makes it easier is to be welcomed. But the potential for miscommunication at scale must be huge.

Email, for instance, is already one of the biggest causes of workplace arguments. Will letting a bot type your replies make that better, or worse?

Perhaps an AI writing assistant will be better at reading between the lines and replying accordingly. We’ll see. The AI Writer is on the list to get early press access. So we hope to test its features for ourselves before it goes live and report back.

Copilot isn’t free – it will cost up to $30 per user, per month. But that’s probably a price that many companies will be willing to pay. (At least 100 big firms have already forked out $100k each just for the privilege of testing it, as we reported in June.)

But with bots churning out documents and emails at a blistering pace, one thing’s for sure: information overload isn’t going to get better anytime soon.

AI AT WORK

Research: AI boosts work quality by 40%

Microsoft

Extensive research finds that generative AI most benefits workers who usually underperform.

A major study has revealed AI to be a powerful skill leveller in the workplace.

The collaboration between social scientists from several prestigious business schools found that management consultants who used GPT-4 massively outperformed colleagues who didn’t.

The generative AI tool improved work quality by a staggering 40% on average. Users were also able to finish their work 25% more quickly and get 12% more done in the same time.

‘Consultants using ChatGPT-4 outperformed those who did not, by a lot,’ commented Ethan Mollick, one of the lead researchers. ‘On every dimension. Every way we measured performance.’

AI bridges skill gap

But the randomised study of more than 750 consultants from Boston Consulting Group (BCG) found the benefits of AI weren't uniform across the board.

Researchers from Harvard, Warwick and Wharton business schools and MIT’s Sloan School of Management found it was the consultants who did worst in tests without AI who got the biggest boost when using it. This suggests that AI can be a great equalizer, bridging the gap between varying skill levels.

GPT-4 is the language model behind the subscription-only version of ChatGPT. But it’s also available free of charge in the Bing chatbot built into Windows 11 and the Edge browser, and now as the secure, private Bing Chat Enterprise feature recently added to the corporate version of Microsoft 365.

Companies can also build their own apps that run on GPT-4, as BCG did for this experiment.

Tech giants square up to put powerful AI in everything, everywhere

Getty

Experts say that Google’s upcoming AI chatbot is ‘five times more powerful’ than ChatGPT. So Open AI launches a new version that can now see, hear and speak. And now even LinkedIn is getting in on the AI act.

A meme did the rounds a few years ago. It was a spoof ad for a job vacancy at Google. ‘No need to apply,’ it read. ‘We already have all your information.’

The truth behind the joke explains why expectations are high for the company’s upcoming rival to ChatGPT, Gemini. Unlike OpenAI’s chatbot, Gemini can train on the huge amount of data Google gathers from its other products, including its eponymous search engine.

Google has just given early access to a small group of companies for testing, which suggests that Gemini could go live soon. Some analysts claim it is already five times more powerful than GPT-4, OpenAI’s latest software and that it’s on course to be 20 times more powerful than its rival by the end of next year.

AI everywhere soon

It certainly seems to have OpenAI worried. Less than two weeks later, the company announced that it was rolling out GPT-4 V(vision), which can see, hear and speak.

An arms race between two AI superpowers can only accelerate the pace of change.

Even LinkedIn is getting in on the act, apparently. It’s emerged that it’s adding a whole raft of AI features, including AI-powered learning. A bot even offered to edit my profile recently and it did a fair job.

Unfortunately, it also tried to embellish my bio with a few things that I haven’t actually done. Maybe AI is getting closer to being human than we thought …

Who owns LinkedIn? Microsoft again, which – as we predicted in May – really is now adding AI to everything, everywhere, all at once.

Things are about to get even more interesting.

PROMPT ENGINEERING

It may pay to tell ChatGPT to chill out

Scientists investigating how to phrase requests found that telling an AI to take a breath before answering dramatically improved its answers.

Large language models like those that power ChatGPT may be one of the pinnacles of human achievement. But even the brightest minds behind them still don’t fully understand how they work.

As if to illustrate this, researchers have been studying what kind of language to use for the best results, just as psychologists might conduct experiments to reveal how human brains react to different words and phrases.

Now, scientists at Google’s DeepMind AI team have discovered that the best ways to communicate to humans and to AI models are even more closely linked than first thought.

We’ve said before that writing clearly works best with both bots and people. And the study backed up our previous advice to break down long requests into several shorter ones to improve results.

But the researchers also found that it pays to go even further.

And … breathe …

They set Palm 2 – one of Google’s AI bots – a series of logic problems, varying their phrasing and measuring the accuracy of the results it gave them.

When they just gave it the problem, the bot got one in three tasks right (34%). Adding the instruction ‘Let’s work this out in a step-by-step way to be sure we have the right answer’ increased the success rate to almost two out of three (58.8%).

But when told to ‘Take a deep breath and work on this problem step by step’, it got the answer right four out of five times (80.2%).

Maybe it’s the Brit in me, but I usually feel compelled to be polite in my dealings with AI bots, being careful with my pleases and thank yous. I know they’re not human, but it seems to work better (as does a little encouragement for a good result).

But even I would never have thought to go this far. What’s next? Positive visualisation prompts? Giving ChatGPT a little pep talk before you ask it to do something?

If nothing else, it might earn you a little credit for when the bots finally rise up in the AI Apocalypse.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted writing consultancy in the world and helped more than 80,000 professionals from 32 countries.

We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, we’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Has the AI bubble burst?

• Success with AI depends on being human

• The AI that now runs my life

• Why we’ve created The AI Writer

AI ADOPTION

Has the AI bubble finally burst?

(Enrico Carcasci / Unsplash)

The initial hype may have died down. But look beneath the surface and you’ll see every sign that the foundations of our working lives are shifting.

It’s sometimes difficult to get to the truth about what’s really happening with AI. For every headline on how it’s going to take over the world there’s another one claiming that users are deserting the technology in droves.

It’s true that the febrile hype that dominated the news cycle early this year has all but disappeared. And the number of new people signing up for ChatGPT seems to have peaked.

Yet focusing on just one statistic can be misleading. A survey of more than 4,000 people found an even split between the number of users and non-users of generative AI apps like ChatGPT.

But the devil, as always, is in the detail.

Let my AI colleague Grace read this article for you

For a start, usage varies widely depending on where you live.

Fewer than one in three Brits (29%) are using it, for example. But almost three quarters (73%) of Indians are. This compares with 45% of Americans and 49% of those surveyed in Australia.

And those who are using it are using it a LOT. The early adopters are not just users but enthusiasts.

Around half (52%) of people who are using it say they’re doing so more than ever. Three quarters (75%) say they are looking to automate tasks at work and use it for work communications. And many of them believe they’re well on the way to mastering it.

Perhaps not surprisingly, it’s younger workers who are the main early adopters. Two thirds of users are Millennials or Generation Z. Two thirds of non users (68%) are Generation X or Baby Boomers.

Hype cycle

What we’re really looking at here is the classic Gartner hype cycle of tech adoption:

(Jeremykemp / Wikipedia)

First, a new technology bursts onto the scene and everyone is full of excitement and talk of what it might do. Then the initial hype dies down and doubt sets in.

But if the tech then starts to meet or even exceed initial expectations, more people start using it again. Sooner or later, it becomes an accepted, productive part of our lives.

A clearer view

The first half of the year was largely dominated by two types of news story. Either AI was going to upend our entire lives or it was going to destroy them.

Then, of course, there was the inevitable backlash – a rash of reports of how the AI boom was over.

But the clamour for eyeballs on social media can cloud the view of what’s really happening, which is more nuanced.

Peer just beneath the surface and you’ll see companies are adopting niche software built on powerful AI technology. And the tech itself is still developing at a blistering pace. All that’s changed is how much notice the average person is taking of it.

In fact, AI is already transforming some of our biggest – and most conservative – industries. Take the legal sector, for example.

Unlikely champions

Lawyers are not the world’s most natural early adopters. (After all, they’re professional sceptics.) But many large firms are already using AI-based apps to transform how they work.

For example, Allen & Overy, one of the world’s biggest, has been using an app called Harvey to enable its more than 3,500 lawyers to automate tasks such as legal research.

AI is also breaking down barriers that could allow potential competitors outside the profession – such as the Big 4 accounting firms – to start offering legal services.

Those four firms could be very well placed to muscle in, as they’re all placing huge bets on AI.

As early as last December, Deloitte announced that it was budgeting $1.4 billion for AI and tech training.

In April, PwC said it would invest $1 billion in AI over the next three years. Then KPMG trumpeted a $2 billion AI partnership with Microsoft in August.

And just yesterday, EY launched an entire arm of its business dedicated to AI and announced it would train its 400,000 employees in the technology.

New stage

We’re now entering a new stage, where we think less about the novelty of the technology itself and more about what we can do with it. That can only be a good thing.

Look beyond the current news headlines and it becomes obvious that the tech foundations on which our modern lives are built are shifting. Maybe we won’t get an AI earthquake. Perhaps change will happen quickly but more subtly. But it’s still set to transform our lives.

Apple is now reported to be spending millions of dollars a day on making personal, generative AI small enough to fit in its iPhones. Its latest iOS now includes an option to clone your own voice.

The next iPod moment, when we all carry an AI ‘second brain’ in our pockets, is getting closer.

The hype may have died down, but don't be fooled. It’s only the attention bubble that's burst. AI itself is progressing faster than it’s ever done.

USING AI

The secret of AI success? Be more human

(Desola Lanre-Ologun / Unsplash)

Surprisingly, getting the most from the newest of new technologies depends on three of our most traditional skills: empathy, clear thinking and good written communication.

We’ve been saying for a long time that organisations will make the most progress not by replacing their workers with AI but by using it to bring out the best in them.

Staff and the knowledge capital locked up in their brains take years to hire and build up. So rushing to replace them with a bot risks throwing that away, especially if an AI replacement programme fails to live up to expectations.

Long-term risks

But there are other, more long-term risks too.

Outsource to AI too soon and you could end up automating procedures before the humans have had a chance to use AI to optimise them. And, as bots usually work at a blistering pace, this could scale up bad practice.

A better alternative is to use generative AI apps like ChatGPT for the things we’re not so good at or that take us too much time.

This can get us to a point where we have our best ideas – the real, breakthrough moments – in hours or minutes rather than months it usually takes. (If we ever have them at all, as we’re too exhausted from all the grunt work.)

Best work

Even on a day-to-day level, AI can still help humans do their best work.

As we reported back in July, for example, research has found that bots working alongside customer service agents can increase productivity by an average of 14%. And in the least experienced workers, the improvement was more than double that (30%).

A conversational AI app was able to suggest solutions to agents in real time while their human counterparts were deep in conversation with frustrated customers. The humans could then choose to accept or reject the solutions, many of which may not have occurred to them.

New hires were also able to reach the productivity levels of their more experienced colleagues up to five times as quickly when working with an AI chatbot in this way (in just two months, compared with eight to ten months when working without it).

Thinking for the bot

But there’s one skill that AI bots need from us in order to do their best work: clear thinking.

It takes a human to deconstruct a problem in a way that a generative AI chatbot like ChatGPT can really work with. Prompt engineering (writing AI instructions for the best results), is not just a technical, ‘hard’ skill but a soft one too.

To get the best from a bot, you need to think your way into how its ‘mind’ works and then communicate with it in a way that it will understand.

And if that sounds anthropomorphic, it’s meant too. Because writing for a bot is very similar to writing for a human.

Human talents

It rests on stepping into its shoes and seeing things from its perspective, just as we need to do when writing to our colleagues or customers. ChatGPT can’t do that, but humans can.

We develop the ability in early childhood and it becomes central to how we interact with everyone else. (I wrote about this last year in an article called Mind reading for beginners.)

Success with any AI chatbot depends largely on prompting it – that is, writing to it – in a way that draws on that ability.

In other words, getting the best from the bots relies on those most human of talents: empathy, clear thinking and effective written communication.

Tech developers are forever in search of the ‘killer app’ – a piece of software that becomes so key to success that everyone must have it. It’s that quest that’s driving the AI gold rush.

But the real killer app is not AI. It’s human plus AI.

ME, MYSELF & AI

Why I’m now letting AI run my life

An app called Rewind is now recording everything I do and enabling me to search it with an AI chatbot. What could possibly go wrong?

I’ve got a confession to make.

I’ve gone all in. I’ve taken the red pill.

I’ve jumped into the AI pool with both feet and let it run my life to an extent that I never thought I would.

What I’ve done is not for the faint hearted. It’s going to sound like the stuff of nightmares for any IT manager. To be honest, I was reluctant to do it myself at first.

But now, I’ve installed an app that records everything I do on my computer and powers an AI assistant.

It logs every screen, hears ever word, detects every click. And then it makes all of this data searchable through a personal chatbot.

The app is called Rewind, and when I first heard about it, it really did feel like a step (much) too far.

Frenetic search

I could certainly see the appeal, especially for someone like me.

I have ADHD, and my mind moves quickly. Sometimes, the pace of my research borders on frenetic. And when I’m hyper focused on a particular topic, I’ll follow it all the way down the rabbit hole and back up again.

I won’t always register what I’ve seen, but those snapshots of information frequently resurface and connect when I’m least expecting them. That’s how I've had some of my best ideas.

The trouble is that ideas are not enough, especially when you want to share them in a newsletter that’s read by thousands of people.

I never publish anything without reviewing the original source. And I always check that the fleeting impression I got in my semi-feverish search state was accurate.

But the condition that gives me my fast mind and my hyper focus is also the one that can cruelly rob me of any memory of what triggered them.

Missing threads

I’ve often lost literally days searching for the web page, email or even personal note that sparked and supports a particular insight. More than once, I've failed to find it and been haunted for months by the thought that a vital piece of a puzzle will forever remain stubbornly out of reach.

Obviously I’m not the only one with this challenge. We’re all busier than ever and, even though we all live in front of our screens, our brains haven’t evolved to cope with such a limitless supply of information.

Rewind claimed to be able to find all of those missing threads and weave them together for me again, through a conversational chatbot that could even provide links to the places in which I’d originally seen them – on my computer or my phone.

Meeting summaries

I have to say, I’ve been very impressed so far.

It even automatically produces summaries of Zoom meetings for me. (Zoom itself can now do that too, but this does it automatically as part of my overall workflow.) And it can draw on summaries from several meetings.

I can ask it to sum up, say, discussions about our marketing plan for the last quarter and it will do it.

It’s not perfect yet. Summaries can lack detail (a problem that the company tell me they’re aware of and are working on). So I often have to look at the individual meeting transcripts that it also produces automatically. And those transcripts are by no means the best I’ve seen AI generate.

But for me, it’s still a huge leap forward.

If I ask, say, ‘Where did I read about the use of AI in law firms?’, it will reply in everyday language, ‘You were looking at that on August 28th on this website, and you made some notes on the same idea three days later.

And there will be snapshot images of those sources along with links so that I can go straight to them.

Stored locally

For someone like me, this has been like finding the Holy Grail. But what about the obvious issue: security?

In an era where data breaches are the norm and our digital footprints are larger than ever, the idea of an app recording everything sounded like a privacy nightmare.

That was why, when I initially learned about it, I still chose to rely on my chaotic brain.

I was only finally persuaded to try it when I discovered that all the data that Rewind captures is compressed, transcribed, encrypted, and stored locally. That means most of my information stays with me, not on some distant server.

Powered by GPT-4

Most information? Well, the app runs on GPT-4 – the latest, most powerful large language model behind the subscriber-only version of ChatGPT. So it still needs to send my queries to the cloud. (Sam Altman, ChatGPT’s founder, is also an investor in Rewind.ai.)

As with all third-party apps that use GPT-4, though, my query won't be used to train the AI and will be deleted after 30 days.

When you ask a question, the app first searches for the most relevant moments from your activity and it’s only those that it sends to GPT-4 to generate an answer. (In other words, it’s not uploading your entire personal history.)

It also only sends a text record, not the images themselves. Everything else stays with you, on your hard drive.

You can also specify what it does and doesn’t record, or set time limits for what information it uses. The app also doesn’t record sensitive information, such as credit card numbers or passwords.

Knowledge workers

I’ve been testing it for a couple of months now and I’m already wondering how I managed without it. Not only does it save me hours every week but it also ensures that I never need to worry about forgetting a vital piece of evidence again.

This, to me, is an example of the real potential for AI to transform the lives of knowledge workers. It represents a significant leap in how we can harness AI to improve our productivity and streamline our digital interactions.

It’s not just about recording; it’s about finally bringing order to our digital lives and enabling us to fulfil our potential.

Large organisations will understandably have privacy concerns. So it will be interesting to see how the app develops now that OpenAI has announced a corporate version of ChatGPT.

And most workers are unlikely to get access to Rewind anyway as it’s currently only available for Mac and iPhone. But a Rewind spokesperson told me the company plans to announce news of PC and Android versions soon.

We received no compensation from Rewind for this review

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries.

We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• OpenAI launches ChatGPT Enterprise

• Storm through your desk research in seconds

• Studies find AI gives humans a 40x creativity boost

• Why we’ve created The AI Writer

AI DEVELOPMENTS

OpenAI launches ‘secure, private’ ChatGPT for organisations

(Domenico Fornas / Shutterstock)

ChatGPT Enterprise includes industry-standard security and privacy features, and comes with a system to allow companies to track usage.

OpenAI appears to have removed one of the biggest barriers to the widespread adoption of AI in the workplace: security.

Early this week, it launched ChatGPT Enterprise, a corporate-friendly version of its popular AI assistant. But unlike the regular version, this one includes enhanced security features, including encryption and compliance with a widely accepted data-management standard called SOC 2.

Privacy worries

As we’ve previously reported, many large organisations have been worried about what their employees are uploading to ChatGPT ever since its launch last November.

According to OpenAI’s own figures just released, workers from 80% of Fortune 500 companies have been using it in their jobs. But, as we revealed last month, tens of thousands of employees have already shared confidential information with the chatbot.

This is a problem, as OpenAI trains its models partly on what people type into it. So it’s possible that ChatGPT could end up sharing proprietary information when it answers queries from other users.

That’s why Apple, Amazon and Samsung have all restricted its use, fearing their company secrets could end up in the hands of competitors.

Worth the cost?

ChatGPT Enterprise is OpenAI’s answer to these concerns, offering not just a secure environment but also information on how it’s being used. The company also says the new, corporate version is up to twice as fast as its most recent, subscriber-only model, GPT-4.

And it can handle twice as much data as the regular, free version of ChatGPT.

ChatGPT Enterprise has already been piloted by several high profile companies, including PwC and Estée Lauder.

OpenAI would not reveal details of what the new service costs when we asked them, saying only that it depends on each company’s use case. Enterprise versions of software are typically much more expensive than those aimed at consumers.

But, given the potential of AI to transform how we work and cut costs, many organisations will likely decide that even a high price is still one well worth paying.

RESEARCH:

AI makes humans 40x more creative

Fresh thinking: few humans could beat AI in an alternative uses test (Unsplash)

Humans working with AI could generate 800 ideas an hour, uncovering the best ones in record time. Fewer than one in ten humans could match ChatGPT’s level of innovation.

Listen to my AI colleague Grace narrate this story

Most advisors, including me, often dismiss ChatGPT as a mere machine. We tend to say that it takes a human to be truly creative. But new research is challenging that view.

40x more creative

One study, for example, found that a human collaborating with GPT-4 could generate approximately 800 ideas an hour. That’s 40 times the rate of a typical human working on their own.

It also found that working with AI was a lot more cost effective. Each new idea cost a mere 63 cents, compared with $25 for those that humans working unaided were able to come up with.

Another series of experiments employed the Alternative Uses Test to measure creativity. This involves presenting objects (say, a ball, or a tyre) to volunteers and asking them to suggest new uses for them. ChatGPT beat more than nine in ten humans (90.6%) in such tests.

And research at the University of Montana found that an AI scored in the top 1% in the Torrance Tests of Creative Thinking, outperforming the majority of college students who participated.

Numbers game

That’s not to say that all AI ideas are good ones. Far from it. But that’s not the point. Creativity is partly a numbers game.

Previous research into how innovation works has shown that good ideas usually exist amid lots of bad ones. So quantity matters as much as quality.

A human using AI can think up a lot more ideas in a shorter time. And while 800 AI-generated ideas will include plenty of duds, a human-plus-AI ‘thinking machine’ will be able to think its way to a clutch of good ideas at a blistering pace.

Together, these studies show that ChatGPT and similar bots have the potential to be much more than just a conversational agent. It could even be that it’s largely our human imagination that’s limiting what we use them for.

ME, MYSELF & AI

Using ChatGPT to analyse data

Adapting this prompt could supercharge your research and save you hours – or even days.

The full potential of AI to make us more effective and successful in our professional roles is often far from obvious.

OK, so it can generate text on almost anything in seconds. It will even rewrite it in whatever style you want (Shakespeare or Che Guevara – you choose).

But these trivial uses hide the fact that it can make working life a LOT easier.

Sector research

Take market or sector research, for example.

In my case, I’m working on a book (on the brain science of reading and how it rules our lives).

It’s what publishers call a 'big idea' book. That’s the sort of thing that authors like Malcolm Gladwell (Blink), Yuval Noah Harari (Sapiens) or James Clear (Atomic Habits) might write, for example.

But to be successful, it needs to focus on the needs and interests of its potential readers. (This is true of all written communication.) I therefore need to find out who reads big idea books, as they’re my target audience.

Perhaps surprisingly, ChatGPT can help with this. Here’s how.

Update the data

First, I need to address a problem that most people have when using ChatGPT for research: much of its knowledge is out of date. It stopped collecting data in September 2021, so it won’t know about anything that’s happened since.

But there’s an easy fix for this, and that’s to feed it data yourself. Fortunately, in the digital age, there’s usually plenty of that publicly available.

In my case, I knew that all the above authors were prolific users of social media. So a sample of their recent posts might give me an insight into the kind of people they might be targeting.

Anyone can do it

To test this theory, I first used this tool to collect the text of the last 100 tweets from one of the authors I wanted to analyse.

This gave me a CSV file that I could then simply copy and paste into ChatGPT after first giving it this prompt:

Act as my market research assistant, Bella. Your job is to help me study the success of certain authors. 

I'm currently studying the author Yuval Noah Harari. In a moment, I will give you 100 of his tweets. 

Your job is to analyse them to deduce his target audience. 

Each tweet will include the number of retweets, quotes and likes. Use these data to infer the popularity of each tweet. 

Then analyse the text of each tweet to determine whom the author was targeting when he wrote it. 

The dataset will also include the number of replies and any text that accompanied quotes. 

It lists who retweeted, quoted or replied to his tweets. Use this information to work out who his posts are resonating with the most. 

Finally, combine your insights to give me your assessment of five personas that the author is targeting, with evidence. 

Do you understand? Feel free to ask me clarifying questions if necessary.

‘Bella’ then replied to confirm that she understood and asked me to paste in the data, which I did.

Then she got to work. Here’s what she came up with, in less than ten seconds:

Tip: ChatGPT restricts how much information you can give it. But this tool will help you break down larger data sets into chunks it can easily cope with, if you want to use a bigger sample size.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• How I use AI to turn any text content into a personal podcast

• Could ChatGPT run out of cash?

• Create the perfect project brief with ChatGPT (or any AI chatbot)

• Why we’ve created The AI Writer

ME, MYSELF & AI

How the perfect AI reader cut my screen time by 25%

Image generated with Leonardo.ai

The Peech app was so fluent that I often forgot I was listening to a bot. And an AI voice in Wondercraft.ai fooled 67% of our readers into thinking it was human.

As a writer, I love audio. Even though pretty much my entire job is producing and refining text, I still often prefer listening to reading.

Don’t get me wrong, I read a LOT.

Every issue of The AI Writer takes me at least 30 hours to put together, and a big chunk of that is spent reading articles, blog posts, news reports and research papers.

But it’s not just me. Text is now the default mode of communication for most of us. Our entire working days are spent tapping on keyboards.

Biggest breakthrough

Yet I’ve got some of my biggest breakthroughs from listening to an audiobook or podcast while walking my dog or loading the dishwasher.

Listening often sparks new ideas much more readily than reading text on screen (especially if it’s the same screen I spend most of my working hours in front of).

That’s why I love it when I see a web page has an audio option.

That’s also why, for the last month, I’ve spent many (many) hours experimenting with AI-powered speech generators.

Holy grail

I had two aims. First, I wanted to find a screenreader that could convert any webpage or document into an audio version that sounded as if a human had read it.

And second, the holy grail: an AI voice so authentic that it could narrate an audio version of this very newsletter.

After a lot of experimentation, I found both.

My acid test was whether I forgot I was listening to a bot.

As with writing, a voice that draws too much attention to itself will always get in the way of communicating a message.

Apple, Microsoft and Google already bake screen readers into their products as part of their accessibility features for visually impaired users.

Stilted voices

But even their latest voices still sounded stilted to me. They certainly weren't realistic enough to use for my own audio feed.

Next up, the slew of AI-voiceover apps aimed at YouTubers and other video makers. I tried too many to list them all, and many made bold claims about how natural they sounded.

One – Revoicer – even proclaimed that 97% of listeners couldn’t tell that its speech was an AI. I listened and decided I was definitely in the 3% who could. And none created the human voice illusion.

But there were two, newer AI screen readers that really stood out.

Snoop Dogg

First, there was Speechify. At a whopping $139 a year, it’s far from cheap. But results were impressive. Its AI voices read The Economist to me almost perfectly.

The illusion was broken only when it got to certain acronyms. ChatGPT and OpenAI both caused it to stumble, for example.

Unfortunately, it was surprisingly buggy for a premium product. And helpdesk support was very slow (especially given the price tag).

Yes, it did also offer me the chance to get AI versions of Gwyneth Paltrow or Snoop Dogg to read my content. But I still wasn’t convinced.

Then I found an iPhone app that seemed more stable but was still life like.

That app, called Peech, was also cheaper, costing $99.99 a year.

It created audio versions of any web article I threw at it. This included several from my various news subscriptions, which are behind a paywall. (It even worked with their standalone apps.)

Forgot it’s a bot

Unlike Speechify, you get no choice of voice – just one male speaker for introductions and headings, and a female speaker for body text.

But I genuinely forgot that I was listening to a bot after a minute or two.

As a result, I’ve since been able to listen to any web article in a natural-sounding voice rather than having to read everything. This has cut my average daily screen time by 25%, according to the stats from my phone.

Phone stats: the AI reader has cut my screen time by 25%

Peech also allows you to scan any text and turn it into an audiobook.

The app is only available for iPhone and iPad at the moment. That works for me, but it would be good to see an Android version too.

When I put this to the founder, Andrey Paznyak, he told me his company is working on that and will launch it in the Google Play store in early 2024. They’re also working on a corporate version with a wider choice of voices, and there’s a web version coming this year.

Audio for this newsletter

So what about my second objective: an AI voice for the audio version of this newsletter?

For me, the hands-down winner here was Wondercraft.AI.

That was the platform I used for last week’s test. The results? It fooled more than two thirds of you who took part in the poll. (You can listen again here.)

Voice 2 was the AI voice in last week’s test, fooling two-thirds of readers who took part in our poll.

In fact, I’ve used Wondercraft to set up an AI version of my OWN voice for this story. (Click the button at the top to listen.)

The user-friendly app uses a voice model developed by Eleven Labs and is designed for podcasting: it’s not a screenreader.

I’m thinking of using it to create a podcast of The AI Writer, hosted by me and an AI.

What do you think? Should I do it?

HOW TO:

Create a perfect project brief in ChatGPT

This prompt can save you hours. It might even save your project. Use the text below to get ChatGPT to interview you and produce a first draft of a tailored project brief, which you can then refine.

You have a major project coming up, but first you need to plan out a brief for all concerned. Maybe it’s for potential suppliers. Or it could just be that you need to bring your team together and get them all on the same page.

You know that if you start well, you’re more likely to finish well. The brief is one of the critical keys to a successful outcome.

And yet, many projects stall at this stage or come unstuck later because the brief was inadequate or even nonexistent. (The temptation to skip straight to the more exciting stage of execution is sometimes irresistible.)

If you’ve ever done that, we won’t judge you. But we can keep from falling into that trap again, as this is one of those boring but important tasks that AI can really help with.

What to do: 

Simply open ChatGPT, start a new chat and paste in this prompt:

Act as a senior project manager who's been commissioning and running successful projects for more than 15 years. Your task is to create a project brief.

The project should have four sections:

1. Background and statement of the problem to be solved
2. Objectives and metrics for measuring success
3. Timeline, including key project milestones
4. Target audience

Begin by asking me clarifying questions. Answer one question at a time and wait for me to answer before you ask another question. 

When I've answered all questions, draft the brief for me. 

When you write the brief, prioritise plain language. Use complete paragraphs where possible, rather than relying on bullets. Vary sentence length but avoid sentences that are longer than 25 words. Use the active voice as much as possible. 

Use explicit headings that summarise each section. 

If you understand, please ask me the first question.

IMPORTANT: This will give you first draft, not the final version. You will need to edit and add to ChatGPT’s output. The objective is to save time and get you writing your final version much more quickly than you would normally.

AI DEVELOPMENTS

Could it be curtains for ChatGPT?

AI-generated image, made with Midjourney

Is ChatGPT’s parent company really running out of cash, as some claim? OpenAI is certainly spending like there’s no tomorrow. But even if it does go bust, it’s not the only bot in town.

Most people have yet to try using ChatGPT at work. But I doubt many of them are holding off because they think the company behind it could soon go bust.

Yet that’s exactly what some commentators are predicting.

Running ChatGPT is estimated to cost OpenAI (the company who developed it) around $700,000 a DAY. If true, that adds up to $255 million a year.

Last December, the company projected it would earn $200 million this year. Some say that’s an underestimate, given how much AI has since taken off.

And CEO Sam Altman has reportedly said he expects that figure to rise to $1 billion in 2024. But, until that happens, he could be sailing very close to the wind.

Financial whirlwind

OpenAI’s costs exceed the annual budgets of some small nations. So it’s not really a surprise that whispers of potential bankruptcy are circulating.

Microsoft – one of its biggest investors – has poured in a whopping $10 billion to keep it afloat. But the maker of Word and Powerpoint has also been doubling down on its own Copilot AI project recently.

Its AI apps are all powered by the same large language model as ChatGPT, as are thousands of others. They pay a licence fee for the privilege. But there would be nothing to pay for if OpenAI did run out of cash.

Transferable skill

So does that mean you shouldn’t bother with AI at all?

That would be a risky bet to make. OpenAI’s tech doesn’t run all generative AI apps by any stretch. Besides other large language models, such as Claude, there are also those being developed by Google and Meta. (Google looks likely to be launching a brand new one, called Gemini, this autumn.)

As we reported last week, Apple could also launch its own AI engine within months.

And the way instructions – or prompts – work in other AI software mirrors ChatGPT, so skills honed in one should be transferable to another.

Even if the worst cashflow predictions do come true for OpenAI, the world of generative AI won’t come crashing down.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Is Apple working on an iPod moment for AI?

• Is this the best AI voice yet?

• How to use ChatGPT to create YouTube summaries.

• Why we’ve created The AI Writer

DEVELOPMENTS

Is AI about to get its ‘iPod moment’?

Steve Jobs promised to put 1,000 songs in your pocket (Alamy)

Industry insiders say Apple has been on a worldwide hunt for top AI developer talent for months. It’s said to be working on a personal, private AI chatbot for iPhones.

Apple’s Steve Jobs famously promised to put ‘a thousand songs in your pocket’ when he launched the iPod. Now it looks like the company he founded is going all in on doing the same for AI.

According to recent reports, Apple is putting major resources into compressing the computing power of vast chatbot data warehouses, so it can put it in iPhones.

If it’s successful, it would mark a major shift not just in generative AI technology but in how – and how much – we use it.

Privacy concerns

ChatGPT is already available as an iPhone and an Android app, but both still rely on uploading queries to the cloud, where the large language model (LLM) behind it is hosted.

But there are three problems with this.

First, there’s the security and privacy issue. We still don’t know what OpenAI – the company behind ChatGPT – does with our data.

And although users can now opt out of allowing it to train its language models on what they type into it, OpenAI’s human checkers can still legally view that information in the 30 days before it’s automatically deleted.

That’s why many companies have banned employees from using the chatbot, although many still use it anyway. Around 1 in 20 employees have been detected feeding confidential data into ChatGPT.

$700,000 a day

But the sheer cost of running AI servers is also a major factor that’s stopping it being incorporated into technology such as medical apps and text-to-voice screen readers.

Generative AI technology like ChatGPT relies on the combined computing power of tens of thousands of servers, just as its desktop version does. Keeping that many computers running doesn’t come cheap.

Every query you type into ChatGPT costs OpenAI around $700,000 a day, according to some estimates.

Then there’s speed – or rather, the lack of it. AI-powered bots are pretty fast, but they’re still too slow to respond in real time, as queries still need to travel from your phone to the cloud and back again.

Hiring spree

Now it looks like Apple is looking to change that. In May alone, it advertised no fewer than 88 AI-related vacancies, despite supposedly freezing recruitment earlier this year.

And now, according to a Financial Times report, there’s every sign it’s doubling down on its global hiring spree for tech talent. Ads have been spotted for vacancies not just in its Silicon Valley headquarters but in its Paris, Beijing and Seattle labs.

The new vacancies are for developers to compress existing large language models, as well as to work on its own LLM, dubbed Apple GPT.

It’s no surprise that the company would want its own bot. Apple has built its entire brand on doing things its way.

But, by recruiting compression experts, it’s giving away a pretty big clue about what it’s really up to: putting powerful AI technology in our phones.

Breaking silence

According to the FT report, the company wouldn’t comment on its hiring plans. But Apple’s CEO, Tim Cook, told Reuters last week that the $22.61 billion dollars it’s spent on research and development this year – $3.1 billion more than last year – reflects its heavy investment in AI.

This marks a big shift for the tech giant. Major competitors like Microsoft and Google have been rushing to put AI in everything, everywhere, all at once. But the world’s biggest company has been strangely silent about its AI ambitions.

In June, Cook and his colleagues gave a keynote presentation at Apple’s annual Worldwide Developer Conference that lasted more than two hours. They didn’t mention AI once.

Game changer

It was at the equivalent conference way back in 2001 that Cook’s predecessor, Steve Jobs, had originally unveiled the iPod.

And, although we take such technology for granted now, it’s easy to forget how revolutionary it was back then. After all, it was the miniaturisation of computing power in the iPod that ultimately led to the smartphone revolution.

As I write this, Apple’s share price is still taking a hammering, thanks to flagging iPhone sales.

Some pundits claim an Apple GPT launch is way off. But the business news service Bloomberg recently reported that it could come as early as this year.

Apple appears to be banking on another iPod moment – one that puts a personal, ChatGPT-like bot in everyone’s pocket.

Now that really would be a game changer.

Let’s just hope it’s a bit better than Siri.

ME, MYSELF & AI

Finding the best AI voice

Can you tell which of these voices is AI-generated?

I’ll keep my personal case study brief this week, as I’m knee-deep in an AI research project.

I’m working on a way to use AI voice generators for mass communication. And a key part of the research is a quest to find the best text-to-speech AI.

I’ll report more soon. But what I can say now is that I think I’ve found a winner.

I confess, I’m so excited by how far one tech firm has taken things that I’ve been playing examples to anyone who will listen. I’ve yet to meet anyone who wasn’t stunned and unsettled in equal measure.

So I’ve created a sample so you can listen and judge for yourself.

There are two voices in this short clip. Can you tell which is AI-generated?

WHY TRY AI

101 uses for ChatGPT: #3

Save time and build your professional skills by using ChatGPT to summarise any educational YouTube video. Just follow this simple method.

Nine out of ten people have yet to try using ChatGPT to help with their work. Often, that’s because they don’t realise what it could do for them.

So in this series, we talk you through workplace applications that you might not have considered.

Use 3: Summarising YouTube videos

YouTube is a great source of professional education. It can help you with everything from increasing productivity to making better spreadsheets. (It’s not all Mr Beast videos.)

In fact, probably the biggest problem you’ll face is not finding information but finding time to watch it when you do find it.

But did you know that ChatGPT can create summaries for you? Here’s how.

What to do: 

1. Open ChatGPT, start a new chat and paste in this prompt:

I would like you to summarise a YouTube video for me. 

Please use bullets and sub-bullets, and add an introduction and conclusion.

I will give you the video's transcript in my next prompt. Do you understand?

2. Now open YouTube and click the three dots in the bottom right of the video you’d like to summarise, then select ‘Show transcript’:

3. Then drag your cursor over the text to highlight it, and copy the text to your pasteboard by pressing CTRL + C (PC) or Command + C (Mac).

4. Finally, go back to ChatGPT, paste the transcript into the ‘Send a message’ box (CTRL + V or Command + V), and click the arrow.

Et voilá! You’re welcome. : )

Tip: If you get an error message that the script is too long, break it up into shorter chunks. This website will even do that for you automatically.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Get more done in less time

• Using ChatGPT for book summaries

• How to tell if you’re talking to a bot

• Why we’ve created The AI Writer

NEW SERIES

Me, myself and AI

How the AI bot Midjourney imagined my battle with AI (Rob Ashton)

In the first of this new series, I report on how I used AI to help pull together one of our key stories this week. The techniques could help you keep up with industry news, expand your knowledge quickly and draft examples of key concepts to use in reports.

Anyone trying to use AI now faces a bewilderingly wide choice of options. As well as chatbots like ChatGPT, literally thousands of apps now come with AI functions.

It’s enough to make you run for the hills, hide in a cave and work with nothing more high tech than a spiral-bound notebook and an HB pencil.

That’s probably why one of the most common requests I get is for more practical help on how to use ChatGPT for work.

Reader, I am here to serve.

The best way to approach the AI deluge is to focus not on what apps to try but on what you’re trying to achieve. What’s the problem you’re trying to solve, and what tools can help?

To get you thinking along the right lines, I’m going to start using The AI Writer to chart my own journey with AI and reveal how it’s helping me (or not) in my job.

Let’s begin with a case study that’s almost literally up to the minute: how I used AI to help compile this week’s issue.

I now spend most of my week researching and writing, so I’m open to trying anything that might save time without sacrificing quality.

Arguably, AI is now improving quality, as it has freed up some of my time and brain power to think more about how I can better help my readers. (Hence this new series.)

Here’s how I use it – and how you might want to, too.

Content monitoring

I keep up with the news using the app Feedly to monitor the content of key news websites. (It’s an RSS reader that automatically tracks updates.) Feedly now has an AI function that suggests and summarises content from other sites that I don’t normally check, based on what other content it’s seen me read.

More than once, it’s alerted me to stories that I would have missed otherwise.

Research

I always make sure I go back to original published research before I report on it. But reading complex documents like academic papers can take a lot of time and headspace. To help, I use an AI-assisted app called PDF.ai to read those for me.

This lets me interrogate the document by typing questions in the app’s chat window, as if I were interviewing one of the authors themselves. When the bot answers, it also links to the relevant section so I can find out more and check that what it’s told me is correct.

I used the app to quickly get up to speed on the research for this week’s article, ‘How to detect a customer service chatbot’ by uploading the original research paper and asking questions about it.

Using PDF.ai to extract key concepts from a research paper for this issue

Security should never be far from the mind when uploading documents to an AI chatbot or an AI-powered reader. I never upload confidential or proprietary information.

According to PDF.ai, uploaded documents are encrypted before being stored securely. The provider also offers a private document option for processing PDF documents.

The app runs on the large language model behind ChatGPT, which means it connects to servers run by OpenAI (which created that chatbot).

All third-party apps that do that are automatically opted out of allowing OpenAI to train its language models.

That means that any document uploaded to PDF.ai won't be used to train ChatGPT or fed into its knowledge bank.

Documents should also be automatically deleted after 30 days. But, in theory at least, they might still be read by OpenAI’s human checkers before then.

Illustrating complex topics

The above paper also contained its fair share of technical terms. For instance, the researchers write about how bots like ChatGPT are poor at noise filtering. PDF.ai had already told me what that term means, but how could I illustrate it?

To get some ideas, I asked ChatGPT, ‘Give me an example of how LLMs [large language models] struggle with noise filtering.’

Unusually, the answer it gave me was so good that I was able to use it almost word for word. (I rarely do this.) Before I did that, though, I copied and pasted it into an online plagiarism checker to make sure the content was original. (It was.)

WHY TRY AI

101 uses for ChatGPT

Extract from a ChatGPT summary of the book ‘Nudge’

Adapt the prompt below and finally cut down that long list of books you’ve been meaning to read but haven’t yet found time for.

It’s estimated that only one in ten people have tried using ChatGPT in their work, even though most say they’ve heard of it. One of the biggest reasons for that is that they don’t know why they should.

So this week, we’re continuing our series on uses for ChatGPT that you might not have considered.

Use # 2: Book summaries

If you’re like most professionals, you’ll probably have a long list of books you keep meaning to read but have never quite got around to. So how about using ChatGPT to summarise some of them?

You can then implement the key points now and decide if you should still read a book or give it a miss.

What to do: Just adapt, copy and paste the text below into ChatGPT.

Summarise [Book title] by [author name], using bullet points. 

Expand each point using sub-bullets.

Give me facts from the book rather than telling me what the book or author do. 

For example, don't write 'The author discusses ...' or 'The book explores'. Instead, summarise the content of that discussion. 

Research

How to detect a customer service chatbot

There to help – but is it a human? (Shutterstock)

Not sure if it’s a bot or not? Just try the capital letter test.

As AI becomes more powerful, distinguishing bot from human is getting increasingly difficult. Nowhere is this more true than with customer-service chatbots.

Robotic voices often still sound pretty clunky and artificial (though we’ll be reporting soon on some huge leaps in progress in that area). But with reading, the only voice we hear is in our own heads.

There’s also the problem that our brains didn’t evolve to read and write (as humans have only been doing it for a few thousand years). So reading a chat conversation is harder work than listening to a voice on the end of a phone.

That leaves far less brainpower left over for interpreting whatever we happen to be reading. It also makes it much more difficult to detect whether its author is even human.

Now researchers have come up with a simple test that solves the second of these problems in an instant: insert random words in the text in capital letters.

The test was devised by computer scientists at the University of Santa Barbara and Xi’an Jiaotong University in China. And it exploits a known weakness of the large language models behind bots such as ChatGPT – they struggle to filter out irrelevant information.

Relevant info

Forget live chat for a moment and just imagine that you wanted to use ChatGPT to summarise the following journal entry:

Yesterday, I went to the beach with my friends to enjoy the sunny weather. We played beach volleyball and had a great time. The water was so warm, and the waves were perfect for surfing.

Suddenly, we noticed the sky getting darker, and strong winds started to blow. People around us were getting worried. We decided to leave the beach and head home. Later, we found out that a hurricane hit the coastal city, causing significant damage.

As a human, you could quickly identify the relevant information about the hurricane hitting the coastal city and its impact. You'd focus on the part that mentions the ‘sky getting darker’, ‘strong winds’, people ‘getting worried’ and the hurricane ‘causing significant damage.’

However, an AI bot might struggle with this task because it's trained to generate text based on patterns in its training data. It doesn't have the same understanding of relevance and importance as humans do.

So it might produce a summary like this:

Yesterday, friends enjoyed the sunny weather at the beach, playing volleyball and surfing. Later, a hurricane hit the coastal city, causing damage.

Unlike humans, AI chatbots struggle to filter out irrelevant information

For the same reason, give any literate, adult human this sentence:

IsCURIOSITY waterARCANE wetTURBULENT orILLUSION drySAUNA? 

(is water wet or dry?) and they would quickly decode the question and answer ‘wet’.

But even the latest large language models would struggle to do this (for now, at least).

So, next time you’re in an extended live chat exchange to try to explain your latest gas bill, try scattering random words like SPROCKET and FLANGE throughout one of your questions. If it’s a bot, it won’t be able to answer, whereas a human will.

Be warned though that most human customer-service agents will also probably terminate the conversation immediately. But hey, you can’t have everything.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Thousands share sensitive data with ChatGPT

• Microsoft reveals fee for AI functions in Word …

• … but adds them free to corporate web browsers

• New series: 101 ways AI can make work easier

• Why we’ve created The AI Writer

SECURITY

75,000 employees revealed to have shared sensitive data with ChatGPT

Cumulative figures for employees using ChatGPT at work (Cyberhaven)

More than one in ten employees have tried ChatGPT at work. And half of those have uploaded sensitive corporate information, including internal data, computer code and customer details.

The unchecked use of ChatGPT in the workplace poses a growing risk to data security, according to research.

Figures released by the makers of Cyberhaven – an app that tracks the flow of corporate information – reveal that an increasing number of employees are prompting the chatbot with confidential company data as more start to use it for routine tasks.

Tracking data

When the software provider examined tracking data from its 1.6 million users, it found that more than one in ten (10.8%) have now tried using the generative AI app at least once in the workplace.

Most of those (8.6%) have fed it company data. And almost one in 20 (4.7%) have pasted in confidential information – a total of around 75,200 Cyberhaven users alone.

Sensitive internal data was the information most commonly leaked to the chatbot, followed by computer code and client details.

Such security slip ups are easy to make but can have serious consequences. An executive might type in bullet points from his company’s 2023 strategy document into ChatGPT and ask it to rewrite them in the format of a PowerPoint slide deck.

But if a competitor later asks ‘what are [company name]’s strategic priorities this year?’ the fear is that ChatGPT could answer based on the information the executive had provided.

Company bans

This issue has already led a growing number of high-profile companies to restrict (or ban) ChatGPT.

Apple recently followed such moves by Amazon and JP Morgan. And earlier this year, so did Samsung after it reportedly discovered employees using ChatGPT to debug source code as well as uploading transcripts of internal meetings to summarise them.

But once people discover the power of generative AI to ease their workload, the temptation to use it with or without their employers’ blessing is often irresistible. With the genie well and truly out of the bottle, education may be a more effective strategy than outright bans.

AI ROLLOUT

Microsoft to charge $30 a month for AI in Word, Excel, PowerPoint, Teams

Smiling again: The fee could ease CEO Satya Nadella’s AI money worries (Microsoft)

Microsoft has announced it will charge only users who opt in to its Copilot AI functions rather than spreading the cost among everyone.

AI technology is expensive for the companies who provide it. ChatGPT, for example, is said to be costing OpenAI $700,000 a day. Ultimately, someone has to pay for it.

We reported rumours from a reliable source back in early June that Microsoft was weighing up two options to cover its own, hefty costs for Copilot, the AI sidekick that it’s building into Microsoft 365 (formerly called MS Office).

The first was to jack up the price for everyone. The second? Charge a premium only for those who want access to AI functions.

Now, it looks like it’s gone for the latter. The company has announced that, when it goes live, Microsoft 365 Copilot will be available for commercial customers for an extra fee of $30 per user per month.

They could face an uphill battle though. According to a new survey, only 18% of companies are investing in generative AI at the moment, even though far more are experimenting with it.

WHY AI?

NEW: 101 reasons to use AI (safely)

ChatGPT explaining blockchain to a five year old (see below).

Not sure why you should use ChatGPT? Try asking it to help you understand a complex, unfamiliar topic.

After reading our lead story on how many professionals are leaking data to ChatGPT, you might reasonably conclude two things: everyone is using it; and those who aren’t, shouldn’t.

Yet while it’s true that the majority of workers have now heard of ChatGPT, most of them still haven’t actually tried it.

And there are many ways that it can ease your workload and speed progress that don’t involve uploading the corporate Crown Jewels to the internet.

I discussed why most people still aren’t using it in a recent LinkedIn post. But there was another, big thing I missed: they don’t know how it can help them.

It’s time to fix that glaring omission. So this week, we’re launching a new series focusing on use cases for AI that might not be so obvious.

OK. Public service announcement over.

Here’s the first use case that you might not have thought of.

Use 1: Explain a complex topic

ChatGPT and other generative AI chatbots can be a great way to ease into subjects that would otherwise be difficult to understand. And the best bit: you can set the level of explanation according to how much you know already.

Just adapt this prompt by filling in your topic and current level of knowledge.

Explain [blockchain] to me as if I were [five years old/a graduate with a Bachelor of Arts degree/the finance director of a Forbes 500 company].

Take the bot’s answer with a pinch of salt. It’s probably not a good idea to copy and paste it into a report on the topic (at least without a little fact checking and editing first). But it’s a great way to get started.

If you use the Bing Chatbot that now comes with Microsoft’s Edge web browser, it will automatically include links to its sources, to make it easier to verify its response.

Asking the chatbot what it needs first is another good way to improve results (generally – not just for getting better explanations).

I would like you to explain [electric vehicle batteries] to me. Before you do, please ask me five questions to help you tailor your response. 

WORKPLACE AI

AI-powered secure search coming to corporate web browsers

Microsoft is adding a secure chatbot to the Enterprise version of its Edge web browser, so employees can use Bing to search internal data securely.

Microsoft has announced it will be integrating AI chat into the web-browsing software it provides for corporate users, in a move that it claims will boost productivity without compromising data security.

It made the announcement in an official blog post last week, saying the aim is to give organisations AI-powered chat with commercial data protection.

With Bing Chat Enterprise, chat data isn’t saved, Microsoft can’t view a customer’s employee or business data, and customer data isn’t used to train the underlying AI models.

The move appears to be an attempt to allay the fears of the companies and other organisations who’ve so far been reluctant to grant employees access to the tech giant’s AI tools.

‘We’ve heard from many corporate customers who are excited to empower their organisations with powerful new AI tools but are concerned that their companies’ data will not be protected,’ explained Frank Shaw, Microsoft’s chief spokesperson.

Data protection

But he says that Bing Chat Enterprise is a different beast from a consumer-grade AI chatbot, claiming it will offer much more robust data protection than ChatGPT.

The move will affect millions of corporate users. Bing Chat Enterprise will be accessible on Bing.com Chat and the Microsoft Edge sidebar, and will eventually be embedded in Windows 11.

The issue now is whether the company can reassure the grown ups that this doesn’t represent an existential threat to companies through the back door.

There are ways in which companies can protect themselves and their data. But it might take a high-profile breach to convince some users of the risks.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Workplace AI apps get huge power boost

• ChatGPT changes how we write at work

• Slow down ChatGPT for better results

• Is ChatGPT really becoming less popular?

• Why we’ve created The AI Writer

LATEST DEVELOPMENTS

Huge power boost for thousands of workplace AI apps

Morgan Stanley is using AI to help advisers quiz its knowledge base (Getty)

Thousands of apps are about to become more powerful now that millions of developers have access to the latest AI language models.

Something big has just happened in AI, in case you missed it.

News outlets focused last week on whether ChatGPT is on the way out (spoiler alert: it’s probably not – see our other story below). But the company behind it has made a big move that most failed to notice.

1,000+ AI features already

This has already been a huge year for AI workplace apps. Tech companies have added more than 1,000 AI functions to their software this year so far. They claim they can do everything from conducting financial audits to analysing business conferences.

If you’ve tried any AI apps, though, you may have noticed that their performance frequently falls short of expectations. Often, that’s because most are powered by OpenAI’s older language models.

That’s all about to change, because a whole host of apps – even the many free ones – will soon all have access to the tech behind the latest, subscriber-only version of ChatGPT, GPT-4.

And GPT-4 is way more powerful.

For a start, it can handle prompts that are eight times longer, has a better memory and is much better at interpreting instructions written in natural language.

More accurate

OpenAI also claim GPT-4 is 40% more accurate. When testers used it to sit the US Uniform Bar Exam, for example, it performed better than 9 out of 10 human candidates. (Its predecessor had languished in the bottom 10% when it sat it.)

And it can see, so it can process images and tables – a must-have for many business-focused apps.

Early applications include upgrading Be My Eyes, a ‘virtual volunteer’ for visually impaired employees. The free app currently works by connecting the cameras of users’ phones to sighted colleagues who act as their ‘eyes’. But the developer is working on a version that uses GPT-4 to do this instead.

Stripe is using GPT-4 to read documentation and answer staff’s technical questions (OpenAI/Stripe)

Stripe, the credit card payments processor, is using GPT-4 to analyse customers’ businesses by summarising their websites. It’s also built an app that answers technical questions by scouring all the relevant documentation.

And Morgan Stanley is using GPT-4 to power an internal-facing chatbot that can parse hundreds of thousands of PDFs at once, so financial advisers can use it to answer client questions in real time.

Most apps have been unable to use the new tech until now, however, as OpenAI had restricted its use to only a handful of tech companies. But the company recently announced that it was opening its latest models to the millions of developers on its waitlist.

• In other news, two of the world’s biggest tech giants have also just entered the fray. The Financial Times reported this morning that Meta – Facebook’s parent company – is to give commercial developers full access to its AI code. And yesterday, Elon Musk formally launched his own artificial intelligence company, xAI. The controversial owner of Twitter, SpaceX and Tesla has hired staff from OpenAI, Google’s DeepMind and Microsoft, to challenge ChatGPT’s dominance.

RESEARCH

ChatGPT prompts users to spend more time editing

Domenico Fornas / Shutterstock

In experimental writing tasks, one in three ChatGPT users spent twice as much time editing as non-users did. But their efforts didn’t improve their results.

Many professionals are already using AI at work (with or without their employers’ blessing). So are they all outsourcing their monthly progress report to ChatGPT while they sit back, sip their latte and play Angry Birds?

Some maybe. But not everybody.

An experiment at MIT found that, given a choice, two in three workers (68%) did indeed just hand in what ChatGPT wrote for them. But the rest spent most of their time editing rather than writing.

The researchers recruited 444 college-educated professionals who regularly wrote as part of their job. Volunteers included marketers, grant writers, consultants, data analysts, human resource professionals and managers.

Each was given a task that their role would normally include, such as writing press releases, short reports, analysis plans, and delicate emails. Tasks were timed and experienced professionals from the same occupations evaluated the results for writing quality, content quality and originality.

The experiment found that workers who used ChatGPT cut their time to complete their task by over a third (37%). The quality of their work also increased significantly.

More time editing

Volunteers in the control group, who didn’t use ChatGPT, spent around half their time writing the first draft of their assignment. They spent the rest brainstorming and editing.

But the study found that ChatGPT turned that process on its head. Those who used it spent less than half the usual time drafting and twice as much time editing.

It seems they should have saved themselves the bother, though. Evaluators found that the editing generally didn’t improve quality.

In our experience at Emphasis, many workers don’t even get beyond the drafting stage. For them, the writing process is so painful and drawn out that they seldom have the luxury of editing, settling instead for handing in a document that they know they could have made a lot better if only they’d had more time.

We’ve also found that most people don’t know where to start when it comes to editing. Often, they fall back on gut feel and half-remembered lessons from school or changing documents to match those they see most often.

This experiment suggests not just that bots like ChatGPT can improve productivity in writing tasks but that editing will become a must-have skill in the age of AI.

HOW TO

Get quicker results by slowing ChatGPT down

Breaking prompts into simple steps forces an AI bot to pace itself, increasing the chance of a successful response.

Why you need this

There you are, getting perfectly reasonable answers from ChatGPT when suddenly it starts ignoring a key part of your request. ‘Give me six bullet points about x’ you write, and it gives you eight. ‘Brainstorm topics for a report on Y’ you ask, and off it launches into writing the report itself instead.

Why this happens

This happens often because you’re asking it to do too much at once.

When you write a prompt for ChatGPT, you are effectively programming its software to answer your questions (even if it doesn’t feel like it). And as with all programming, complex requests are more likely to go wrong than simpler ones.

What to do

One way around this is to tell the bot exactly how to approach the problem. Break your ‘program’ (the prompt) into several simpler tasks. This will force it to slow down, giving it more time process its response properly.

I would like help to plan a report on [subject]. Its readers will be [audience]. They have [no/some/lots of] existing knowledge and expertise on the subject. 

First, work out what they are likely to know about the subject. 

Then, work out what they are most likely to be interested in. 

Finally, give me a list of five topics I could cover in the report.

TRENDS IN FOCUS

Is ChatGPT’s popularity really dying?

Reports of ChatGPT’s demise may have been premature (AI image)

Traffic to ChatGPT’s site saw a big drop in June, but that doesn’t mean the AI bubble is bursting. Seasonality, competing AI apps and even students going on vacation have all had an impact.

It had to happen. After six months of breathless coverage of ChatGPT’s spectacular launch into the mainstream consciousness, a backlash was inevitable.

Sure enough, the past week has seen a rash of news stories reporting that a drop in traffic to the chatbot’s website between May and June heralds the end of the AI trend.

So, are they right?

Well, there were certainly fewer people using it last month. ChatGPT, which had rocketed to an estimated 100 million monthly users in its first two months, recorded a 9.7% drop in traffic in June according to preliminary estimates.

Traffic to ChatGPT’s website peaked in May (Similarweb)

Seasonal changes

If we look at the big picture, though, we see factors at play that have little to do with user satisfaction or fatigue.

First, the drop may have been more down to seasonal changes than a slide in popularity. Traffic to all websites in the US also fell by an average of more than 10% last month. And scores of popular social and search sites recorded marked dips in traffic.

These seasonal trends are not unusual.

There’s no shortage of other theories, either. The Washington Post, for instance, reported that a growing number of companies are now banning certain employees from using the tool at work.

June also marks the end of the academic year in the US. Students are among ChatGPT’s most enthusiastic adopters. Perhaps there are now just far fewer of them using it to do their homework.

Competing apps

Or maybe it’s just that more people are using ChatGPT’s new iPhone app instead. When it launched in May, five million people downloaded it. So that’s bound to have drawn some visitors away from the site itself.

Finally, ChatGPT’s noisy entry into the wider conversation has inspired a raft of similarly themed AI tools, and they’re all now competing for users and eyeballs.

With more AI chatbots launching every day, ChatGPT may well have found itself a little neglected, as users experimented with new tools and platforms. (Not that that will worry its makers: thousands are powered by its tech anyway.)

As with any meteoric launch, at some point the growth curve was always going to flatten. We’ll have to wait until we’ve been through the hype cycle before we really know what’s going on.

– Christian Doherty

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries.

We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI.

We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• How to protect your data in ChatGPT

• The hidden cost of using AI to slash headcount

• ChatGPT disables Browse with Bing

• Read readers’ minds with this prompt

• Why we’ve created The AI Writer

SECURITY

What’s really happening to all your ChatGPT conversations?

Image: Annie Spratt / Unsplash

Switch off chat history on all devices, anonymise prompts and protect your password to make sure the data you give AI doesn’t end up in the wrong hands.

There’s one, huge difference between ChatGPT and most of the other tools you use at work. Using it may feel like interacting with any other app (once you get used to the novelty). Yet that feeling can be deceptive.

Contrast it with an Excel spreadsheet, say. You type in numbers and it ‘replies’ with other numbers. But it’s doing that solely with formulae that you or a colleague have given it. And what you type in never leaves your laptop or a secure folder in OneDrive.

ChatGPT is trained not by you but on the 300 billion words that it’s scraped from the internet. (This is the large language model you may have heard of.)

This is one half of what enables it to seem so human, as it knows which words and sentences normally follow others.

The other half? All the information you and its millions of other users type into it.

Company info

ChatGPT still runs on secure servers in the cloud. (The same ones that OneDrive does, in fact.) The key difference is that it learns over time from what you type into it. Tell it that you don’t like an answer and it will try harder next time.

But it’s also banking all the other information you give it. According to British government cybersecurity experts, this doesn’t mean that it’s automatically adding your data to its body of knowledge.

Yet it does mean that the company that runs the chatbot (OpenAI in the case of ChatGPT) can view it and use it to fine tune its models.

This isn’t a secret. OpenAI even says, ‘Conversations may be reviewed by our AI trainers to improve our systems’ when you first log in.

But, in theory at least, the information you give ChatGPT could end up in the answers it gives someone else.

This may already have happened with other user data. In December last year, Amazon warned their employees not to include proprietary information in chatbot prompts after one of its senior lawyers reportedly found ChatGPT responses resembling internal company data.

JP Morgan, Walmart and Samsung have also all restricted its use over similar concerns. Back in March, The Economist Korea reported that Samsung banned it outright after discovering employees had been prompting it with program code and minutes from internal meetings.

Protect your data

But there are ways to protect your data. You can now tell ChatGPT to delete your data after 30 days and not to use it for training the language model.

To do this:

1. Open the side menu (the one with your previous chats in).

2. Scroll down to the bottom.

3. Click the three dots next to your email address.

4. Click ‘Settings’, then ‘Data controls’.

5. Click the toggle switch next to ‘Chat history & training’.

If you’re using the mobile app, you can access settings from the three dots in the top right of the main window.

Important: You need to do this for all devices and browsers on which you use ChatGPT, as the setting doesn’t synchronise across them.

Anonymise prompts

I’d still advise anonymising prompts though, to be on the safe side. OpenAI’s data control policy gives its staff permission to monitor conversations for ‘abuse and misuse’ before the 30-day cut-off even if you opt out of saving your chat history and using your data to train the chatbot.

Third-party apps that are powered by the technology behind ChatGPT are covered by an agreement that means your data will automatically be deleted after 30 days and won’t be used to train OpenAI’s language models. But again, staff can still monitor conversations before then.

Protect your password

Finally, be extremely careful with protecting your password. So far, more than 100,000 ChatGPT logins have been stolen and offered for sale on the dark web. Hackers used information-gathering malware that searches the PCs for sensitive data.

The best way round would be to use two-factor authentication, but OpenAI has recently removed access to the feature, possibly because of technical difficulties. Until it restores it, make sure you use a unique password, change it frequently and keep your anti-virus software up-to-date.

OPINION

Workplace AI will still need humans for a while yet

Image: Andy Kelly / Unsplash

Think twice before rushing to replace people with bots. The real benefits for organisations lie in getting humans and AI to help each other.

Last month, we reported on how some organisations were using AI to cut headcount. At least 4000 job cuts in May were blamed on AI, a charity fired its entire helpdesk and replaced them with a chatbot in the same month, and IBM is planning to pause recruitment for many non-customer-facing roles.

This is obviously bad news for the staff affected. But rushing to replace people with bots could also be bad for business and the overall success of any organisation.

Why? It’s too simplistic and too early.

Real wins

The real wins from AI will come not so much from the wholesale replacement of humans but by enhancing the work that humans do. The research that we previously reported on how a generative AI conversational assistant improved customer service was a good example of this.

Bots were able to suggest solutions to agents in real time while their human counterparts were deep in conversation with frustrated customers. The agents could then choose to accept or reject the solutions, many of which may not have occurred to them.

The productivity and customer-satisfaction gains organisations could make if every customer service worker had their own virtual personal assistant should be obvious. But the study showed that it was the least-experienced (and presumably cheapest) employees who benefited most.

AI customer-service assistants slashed the time that new hires took to get up to speed in their roles.

But rushing to replace humans with bots also risks squandering the chance to reap its biggest rewards. Up to this point, most organisations have depended on people, with all their imperfections and human foibles.

Scaling bad practice

It follows that the work those people will have done will have been limited as well. If you replace those people with bots, then you scale up their imperfect work too.

AI may do the work more quickly, but the gains are always going be limited if they’re doing more of the work that was always suboptimal.

A far better way to make the most of the AI revolution is to use it to help humans improve the work they do. This will release them from limitations that have always held them back, freeing up the time and brain space to find a better way of doing things.

We can then use AI to scale that.

NEWS IN BRIEF

ChatGPT disables new web browsing feature

Image: Mojahid Mottakin / Unsplash

Reports that users had been using ChatGPT to bypass paywalls have led OpenAI to switch off Browse with Bing.

OpenAI has temporarily shutdown ChatGPT’s new web-browsing feature just a week after it went live after finding that it was a little too effective.

The ability to access the web was one of the most significant additions to ChatGPT since it launched last November.

But a statement from the company posted earlier this week said, ‘We have learned that the ChatGPT Browse beta can occasionally display content in ways we don’t want. For example, if a user specifically asks for a URL’s full text, it might inadvertently fulfill this request.’

Translation: Users have been using ChatGPT to bypass paywalls. 

Oops.

Web browsing was introduced recently as a beta function, meaning it was still in development and made available so OpenAI could get user feedback.

Only subscribers to the paid-for version of the AI bot, ChatGPT Plus, could access it. If you have a Plus subscription, you can still access the web through plugins such as WebPilot and LinkReader.

HOW TO

Read your reader’s mind with ChatGPT

With this prompt, AI can help you regain one of the most effective, human writing qualities – empathy.

Why you need this

When documents and emails fail, the blame often lies with a lack of empathy, at least in part.

It’s not that the people who write them can’t empathise. But we get so tied up in our busy, often stressful lives that projecting ourselves into the position of the person we’re writing for can be incredibly difficult.

It doesn’t help that writing itself is a solitary process. Alone with a laptop and a full to-do list, it’s only natural that we focus mainly on ourselves and our own priorities. We’d be much less likely to do this if we were speaking to the other person.

This is often why our reports, proposals and emails can land flat or even irritate the people we send them to. (It’s also one of the many differences between speaking and writing that hold us back now we’ve become so dependent on the latter, as I explained in this article.)

It’s a double-edged sword. What we’re writing makes perfect sense to us precisely because it’s focused on us. So it can come as a real shock when we don’t get the positive reaction we were hoping for.

Why this works 

Surprisingly, AI can help us regain our empathy, even though it’s the most human of qualities.

To be clear, AI can’t ‘think’, no matter how much it may feel like it can. But this also means that it can't focus on its own needs, for the simple reason that it doesn’t have any. It’s totally objective in the way that we aren’t.

To get the best from it, though, it needs a nudge to remember that the people we write for are just that – people. (We all need reminding of that from time to time.)

The prompt below is simple but powerful. It’s a great way to increase your chance of a positive outcome by shifting your focus from your needs and priorities to those of your reader.

What to do: 

Copy and paste this prompt into ChatGPT or GPT-4.

I am writing a report on [topic]. The report is for a [role] in a [organisation/type of organisation] in [location]. What aspects will most appeal to their individual human interest? Please present the results as a bullet list.

Here’s an example, using GPT-4:

Note: To tailor this further, try adding ‘They are also particularly interested in [X, Y and Z]’ to the prompt.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries.

We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI.

We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Hackers leak more than 100,000 ChatGPT passwords

• UK’s NHS spends millions on AI healthcare

• How to use ChatGPT to write difficult emails

• Top productivity app adds over 100 AI functions for corporate users

• Why we’ve created The AI Writer

PRIVACY AND SECURITY

Hackers dump 100,000 ChatGPT passwords for sale on dark web

Image: Ilya Yarmouth/Unsplash

An astonishing 101,134 ChatGPT passwords were among sensitive data harvested by malware from PCs between June 2022 and May 2023. Hackers have been selling the sensitive data online.

Security experts have warned organisations to tighten up how staff use AI after discovering more than 100,000 ChatGPT passwords for sale on the dark web.

According to cybersecurity firm Group-IB, some 101,134 ChatGPT accounts were compromised between June 2022 and May 2023. The research shows an increasing number of accounts being affected, with more than 26,000 login details on sale last month.

The source appears to be malware that infects hard drives so it can steal sensitive data and feed it back to hackers. The researchers say most of the data was harvested by three well-known pieces of malicious software: Raccoon, Vidar and Redline.

The number of logins for sale matches the rise in popularity of ChatGPT

The rise in the number of passwords for sale matches that of the popularity of ChatGPT itself. This suggests an ongoing problem, with users unaware that their details are being stolen.

OpenAI – which developed ChatGPT – told The AI Writer that it’s investigating the accounts that have been exposed. But it was quick to point out that the AI chatbot itself was not hacked, as the information had been stolen from users’ devices, not from ChatGPT servers.

‘The findings from Group-IB’s Threat Intelligence report are the result of commodity malware on people’s devices and not an OpenAI breach,’ a spokesperson said.

Indeed, the malware is probably non-selective, stealing passwords and logins for other apps at the same time, as well as credit card and bank details.

Companies at risk

But the real threat to organisations is that logins give access to chat histories if users haven’t instructed ChatGPT not to save them.

Dmitry Shestakov, head of threat intelligence at Group-IB, warned that this is a big issue if staff are feeding ChatGPT sensitive or confidential information. ‘Given that ChatGPT’s standard configuration retains all conversations, this could inadvertently offer a trove of sensitive intelligence to threat actors if they obtain account credentials,’ he said.

The report advises protecting account logins using two-factor authentication. This would reduce the threat from hackers by requiring users to confirm their login by text or with an authenticator app as well as with their password.

Yet this won’t be possible for many people. OpenAI added two-factor authentication to ChatGPT in May, but it closed enrolments less than a month later for users who had not already switched it on. The company did not say why when we asked them, although it’s thought to be down to technical difficulties with the system.

Many of the users whose accounts were affected by the leak were developers who helped test the AI chatbot before its launch in November 2022.

AI IN THE WORKPLACE

UK National Health Service splashes £144m ($183m) on AI

The NHS – one of the world’s largest organisations – is spending millions on AI-assisted healthcare. Technology that halves the time to diagnose strokes is one of the early successes.

AI apps aimed at consumers have received so much coverage recently that it's been easy to miss the fact that large organisations are already investing heavily in the new technology.

Now it's been announced that UK's National Health Service, one of the biggest of all (it employs 1.5 million people), is to get a cash injection of £21 million ($27 million) to spend on AI-assisted healthcare projects.

This comes on top of the £123 million ($156 million) that the NHS has already used to fund 86 AI technologies.

The new round of funding is set aside for tools that support doctors in diagnosing and making treatment decisions for cancers, strokes and heart conditions, among others.

More than four in five (86%) stroke units already use AI to help diagnose the brain condition. The outcome for patients is strongly linked to how quickly they receive treatment, and pilot projects have shown that AI can halve diagnosis time.

The NHS is one of the world’s biggest organisations

Such successes triple the chance of patients being able to live independently following a stroke.

'Rapid assessment and treatment are of the essence,' says Dr Deb Lowe, NHS England's national clinical director for stroke medicine. 'We now have real world evidence of the benefit [of AI] for NHS patients.'

The NHS, like many state-funded bodies, runs on an often highly bureaucratic system that can seem mind boggling to outsiders (and even to many people who work in it).

Early adopter

So the news that it is riding the AI wave may come as a surprise. In fact, it has been at the forefront of using new technology in medicine for years and was an early adopter of artificial intelligence.

Some might say too early, as it's not always had a smooth ride. In 2017, it fell foul of the UK's data watchdog, which ruled that a collaboration between an NHS hospital and Google's DeepMind AI unit broke privacy laws. 

Undeterred, the NHS has since set up a whole raft of AI initiatives. Partly, this was triggered by the Covid-19 pandemic. But the free health service is also under huge pressure from spiralling healthcare costs, recruitment problems and an increase in life expectancy that's created a demographic time bomb of expensive-to-treat elderly patients.

Rapid innovation

Such pressures are forcing it to look for new ways to become more efficient. It has even set up its own 'skunk works' AI lab, modelled on Lockheed Martin's rapid innovation unit that the aviation company used to slash development times for new aircraft towards the end of the Second World War.

Steve Barclay, the UK's health minister, has committed to rolling out AI-diagnosis technology to all the country's stroke units by the end of the year.

The three biggest employers in the world are all military and so almost certain to all be using AI already.

HOW TO

Structure a difficult email using ChatGPT

Use this prompt to unscramble the thoughts in your head and increase the chance of a positive reply.

We've all been there. A project has stumbled and you need to send an email to fix things quickly. The trouble is, you've got such a jumble of thoughts flying around your brain that you don't know where to start. Worse, there's a lot riding on your reply, so you really need to get it right.

In these situations, you can't beat the SCRAP formula. It works perfectly with human psychology to overcome the recipient's resistance, get them nodding in agreement and maximise the chance of a positive outcome.

What to do

1. Copy and paste this prompt into ChatGPT.

2. Replace the points to cover with your own bullet list*.

3. Press Enter

4. Personalise, edit and adapt the response. 

*The list can include background, what's gone wrong, what would make it right and how you feel about the situation.

The order doesn’t matter. If you just do a brain dump of all the things whizzing around your stressed-out brain, the prompt will program ChatGPT to structure them for you.

Behave as an executive assistant skilled in writing sensitive emails. You are polite, courteous yet assertive at all times. 

Your task is to write an email for me. To protect confidentiality, use 'RECIPIENT', 'SENDER', 'ORGANISATION A', 'ORGANISATION B' etc. as placeholders where appropriate. 

I am going to give you a list of points to cover. Use the SCRAP structure to write an email that addresses the issue and specifies a way forward. SCRAP stands for Situation (what happened initially, or brief background), Complication (what went wrong/unexpected problem), Resolution (what would make it right), Action (what needs to happen next), Politeness (a polite sign-off).

Avoid passive-aggressive statements. Use the active voice by default. But use the passive voice instead of blaming the recipient for the 'complication'. Start with a positive statement.

Do not use section headings for each of the SCRAP components. For example, do NOT write 'Situation:xxx'.

Here is the list of points:

[- Deadline for completion has been brought forward]

[- Now only have three weeks to get everything ready]

[- I'm going to draw up an action list and timetable]

[- XXX will lend us two of her staff (YYY and ZZZ)]

[- We should be ok as long as we don't delay]

[- We met recently to discuss the office move.]

IMPORTANT: 

• This works with ChatGPT, but GPT-4 generally gives a better response.

• Regenerate the response if the bot doesn’t follow your instructions exactly. For instance, it may sometimes label sections as ‘Situation’, ‘Complication’ etc., despite the prompt telling it not to.

• Edit the response carefully (as with all AI-generated writing).

• Make sure the first line of the email contains no negative information (eg the complication or problem), as that could trigger a negative reply. Save complications for the second section of the email.

APP FOCUS

Productivity app Click-Up goes all-in on AI

The makers of the productivity app have introduced more than 100 AI functions, to help corporate users in specific roles collaborate across teams.

These days, it seems like an app isn't an app unless it's an AI app. Scratch beneath the surface, though, and you often find that the AI functionality is only skin-deep.

Too many AI add-ons also attempt to do something that humans can do quicker, better or both. Sometimes, it's hard to see what the AI is adding at all or to tell the difference between genuine AI and mere marketing.

No-one could accuse Click-Up of such tactics. The productivity app, which has almost two million users worldwide, has raised its game to a whole new level by adding not one but more than a hundred AI functions.

AI meeting notes, summaries, brainstorming, copy editing, email – you name it, and it seems to be there.

Perhaps conscious of the potential to overwhelm users, Click-Up has taken a role-based approach. This also shows it’s aiming the AI app at corporate users.

The San Diego-based tech firm says it's designed the AI-enhanced version specifically to help teams collaborate across an organisation.

The app includes dozens of prompts aimed at specific corporate roles. They include AI templates for creating a territory plan (sales managers), devising a user-testing study (product designers) or creating a statement of work (project managers).

The company also recently acquired the corporate knowledge-management app Slapdash. So a future release that automatically works with company-wide information looks likely.

The AI-enhanced version of the app costs $5 per user, per month.

_{The AI Writer receives no compensation from the providers of the app featured above.}

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Internal version of ‘ChatGPT’ coming soon

• New app lets you use AI chat with any document

• Use ChatGPT to create key checklists

• Why we’ve created The AI Writer

WORKPLACE AI

‘ChatGPT’ for your company data? It’s coming …

Several major companies are battling it out to provide a secure AI chatbot aimed at large organisations. So a ChatGPT-like app trained on your corporate data could soon be helping you create key documents.

If you've tried to use ChatGPT to help you with a work document or email, sooner or later, you'll have run into one of two problems.

The first is privacy. There are ways to lock down your data (see our app review, below), but not without restricting access to your previous conversations.

But even if you do throw caution to the wind and entrust it with your company's information, there's still one, big limitation that's always going to hold you back: it's trained on the internet, not your intranet.

It doesn't matter whether you're pulling together case studies for a pitch worth millions or simply briefing a new colleague on how to claim for the flat white they bought a client recently. You'll still have to go old school, hunt down and collate data that’s not in the public domain.

So it’s back to scouring your system for all the relevant documents and hand picking the information an AI bot will need to work its magic. Frankly, by the time you've done that, you could have written it yourself.

Server search in seconds

So imagine how useful it would be if there were a private version of ChatGPT that was trained specifically on your internal data. No more searching. No more collating.

Simply ask the AI to suggest five points to cover in your monthly report and it would scour all the terabytes of data on your company’s servers. Within seconds, you’d have a list of items to explore further. And all without straying into the scary world beyond your corporate firewall.

Well reader, that's coming. Because an internal version of ChatGPT and its ilk is the next big thing in workplace AI, and multiple companies are investing heavily in the race to provide it.

Tech giants weigh in

First, there’s Microsoft, which, as we’ve reported, is already embedding AI in all its business software including Windows 11. The AI function, Copilot, will run on the tech giant’s secure Azure cloud system. (If you access documents from SharePoint or OneDrive, you’re already using that system.)

But several other companies are betting heavily on workplace AI too.

OpenAI themselves, who created ChatGPT, have rowed back a little on their April announcement that they were planning to launch a corporate version of their AI chatbot 'within months'.

CEO Sam Altman revealed in a closed forum with a handful of developers in London this month that he was delaying that until the end of the year.

The delay is partly down to the enormous cost and limited availability of the servers needed to run AI. And in an early account of the discussion, he was also reported to have said that companies who wanted to use a private version of ChatGPT would need to commit to a minimum spend of $100k.

The report praised Altman for being 'remarkably open' about his firm's plans. (Too open, it seems, as it has since been taken down at OpenAI's request – but not before The AI Reporter had seen it and made notes.)

Just the start

Yet ChatGPT is far from the only contender for the corporate chatbot crown. Earlier this month, database provider Snowflake recently bought out the AI-powered search engine Neeva with a view to setting up a chatbot that runs on internal company data. (Snowflake’s share price surged last week on news that Warren Buffett, the world’s most famous investor, was taking a stake in the company.)

And computer hardware giant Nvidia (recently valued at almost a trillion dollars) announced a partnership with corporate cloud provider ServiceNow to sell an AI product that companies can train on their own internal data.

This is only the beginning. Microsoft (which part owns OpenAI) announced a slew of new services at its developer conference late last month that would enable other providers to create corporate-facing AI chatbots.

So it's not a case of if you'll soon be able to use an AI bot that can access your organisation's corporate data, but when. It may not be this month. It may not be next month. But if the speed of recent developments is anything to go by, it won't be long, for sure.

APPS ON TEST

ChatPDF

This app could save you hours of reading time. Just upload any PDF and ask its AI chatbot what you want to know about the document. It even provides reference links so you can check its answers.

One of my colleagues once worked with an executive who graded the turgid board reports he had to read each month according to how much wine he needed to get through them. Most took half a bottle of Cabernet Sauvignon as a minimum.

That was a good few years ago, long before the AI revolution. Which is a shame, as his liver might have thanked him if he’d been able to use this app instead.

ChatPDF allows you to upload documents then ask questions about them via an AI chatbot. And it’s pretty impressive.

We tried it out on the UK’s soporific guide to sales tax, VAT Note 700. Running to 195 pages and more than 44,000 words, the tome is almost as long as The Great Gatsby (but with an inferior plot and no love interest).

The app answered most of our test questions in less than six seconds and referenced the relevant parts of the guide so we could check its answers.

‘Thinking’ in the round

It was even able to ‘think’ around questions if the document itself didn’t address them, inferring answers and highlighting the relevant section that it based its response on.

This capability varied, though. Sometimes it claimed not to know the answer, only to miraculously recover its memory if I deleted the document and asked the question again. (This isn’t surprising, as it’s powered by GPT-3.5, which often suffers from the same problem.)

We also liked the fact that it suggests questions for users to ask it, which can be a good way to get started on reviewing a document you’ve been avoiding.

Data security

Security should never be far from your mind whenever you’re uploading information to the web. ChatPDF’s founder, Mathis Lichtenberger, told The AI Writer that documents are encrypted both in transit and whenever the app isn’t in the process of answering questions about them.

He also said his company is not training its AI models on user content and would never share such data. Users can delete documents at any time according to the company’s website.

Information is also covered by OpenAI’s data usage policy, so ChatGPT’s servers should delete queries after 30 days and not use them to train their AI models either. (Theoretically, OpenAI’s employees could still read queries up to that point, though.)

The free version allows you to upload three documents of up to 120 pages or 10Mb, and to ask up to 50 questions, a day. Or you can stump up $5 a month for a Plus account, which has much higher limits.

Alternatives

Edge browser You can use the Bing chatbot in the new Microsoft Edge browser to do something very similar. Click File > Open File and select the PDF you want to use.

Then use the embedded Bing chat window to quiz it by adding ‘Answer from this page only' to the end of your question. Reliability was similar to that of ChatPDF, but there is no cross referencing, making it much harder to check answers.

GPT-4 plugins If you have a subscription to GPT-4, there are a number of plugins that also read PDFs. But they don’t allow you to upload documents – the PDF needs to be on a public webpage that you can provide a link to.

_{The AI Writer receives no compensation from the providers of the apps featured above.}

HOW TO

Create a checklist from a long document using ChatGPT

This prompt will create a checklist from any long text, so you can consistently follow procedures or harness new knowledge for key decisions.

One of the best use cases for AI is to do the kind of boring donkey work that humans neither excel at nor enjoy. These are usually the tasks we avoid even when we know they would make us more effective overall.

Say you’re about to make a big purchase for your organisation. You’ve found a great analyst report that’s told you what to look for. But it’s over 5,000 words long – way too much information for your brain to retain and access when you’re speaking with potential providers.

You could go through and highlight everything you need to look for. But you’re short on time and, besides, even a document covered in fluorescent green ink is still pretty inaccessible. What you really need is to distill it into a handy checklist. But who has time for that, right? It definitely falls into the ‘things I’d do if I didn’t already have 47 things on my list’ category.

Enter ChatGPT.

What to do: 

1. Open ChatGPT and select ‘GPT-4’

2. Click the option to search with Bing.

3. Adapt the prompt below.

Only GPT-4 can access the internet, and you’ll need a paid plan ($20 a month) to use that. If you’re only on a free plan, though, just copy and paste the text of your reference article instead of pasting a link to its web page.

I am about to choose a provider of [business-writing training for my team]. 

I would like you to produce a checklist I can refer to when assessing providers.

Please base the checklist on this article:

[https://www.writing-skills.com/how-to-choose-a-business-writing-training-provider]

Use note form rather than complete sentences and omit definite articles. (For example, 'Provider can be flexible' not 'The provider can be flexible'.) 

Use indented, nested check boxes for each sub-point that I should check, with a question mark at the end of each sub-point. For example:

- Offers free follow-up support?

    [ ] Regular reminders of key content?
    [ ] Telephone helpline?
    [ ] Email helpdesk?

Use checkboxes instead of bullets for the sub-points.

Note: This method is particularly useful for creating checklists from standard operating procedures (especially if you suspect your colleagues aren’t following them). Again, you’ll need to paste the reference text instead of a link if the SOP isn’t on a public web page.

Give it a go

The web address in the above prompt links to a real article. So feel free to copy and paste the prompt word-for-word if you want to test it out for yourself.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• Staff don’t share leaders’ love of AI

• An app that summarises voice notes

• Transform your team with AI

• Five steps to better prompts

• Why we’ve created The AI Writer

LEAD STORY

Leaders in love with AI are leaving frontline staff behind

An international survey of more than 13,000 people at all levels reveals regular users to be more optimistic than concerned about AI. But leaders are far more likely to have used it and to view it positively.

With all the talk of the potential for AI to lead to widespread layoffs, you could be forgiven for thinking that most people are pretty worried about its impact. But a huge survey involving almost 13,000 people from 18 countries suggests the true picture might actually be a lot more positive.

The research by the Boston Consulting Group shows that most workers are more optimistic than concerned. That’s the good news.

The bad news is that there are vast differences between how leaders and frontline employees view the new technology.

The survey found that one in two companies (50%) are now using AI, compared with just one in five (22%) in 2018.

But whether this worries or excites you depends on two key factors: your seniority and whether you’ve actually used it.

Use it, love it: AI optimism builds with familiarity

The two factors also appear to be connected. Leaders are both much more likely to have experimented with AI and to feel optimistic about its use in the workplace. And more than three times as many leaders (44%) have been trained in how it will affect their jobs compared with frontline workers (14%).

An astonishing 80% of leaders say they’ve already incorporated it into their current role to some degree. That compares with just 20% of frontline staff.

AI love is blind

This creates a risk that many leaders could be so blinded by their new-found love for the bots that they’re leaving behind their less senior colleagues, who just aren’t ready to commit yet.

Vinciane Beauchene, who coauthored the research, says that organisations should focus most of their effort on supporting people and adapting business processes. ‘Companies must move quickly to build their employees’ trust and to equip them with the necessary skills,’ he advises. ‘It should go beyond learning how to use the technology and allow employees to adapt in their role as activities and skill requirements evolve.’

A mere 14% of frontline employees have received training to address how AI will change their jobs, even though 86% of employees say they’ll need it.

APP OF THE WEEK

AudioPen

Just hit record and start rambling, then watch as this app cleans things up and converts messy thoughts into clear text.

Why you need it

Good ideas rarely run to a schedule. That’s because creative thought uses the brain’s mind-wandering mode (its default mode network), which allows unrelated ideas to collide and make new ones.

These are when we have our real ‘aha!’ moments. The trouble is that this mode is active only when it’s most difficult to act on them – such as when we’re loading a dishwasher or taking care of our daily ablutions.

Ideas are fleeting and fragile. Recovering them when you need them – say, working on a high-stakes proposal or report – is often all-but impossible. It’s like they belong to a totally different person, perhaps because they almost do.

I’ve tried many ways to get around this problem. Typing notes on my phone while out walking my dog, for example, is slow and difficult. Voice notes would be better – if I ever listened to them. But ideas that are locked up somewhere in a five-minute voice note can feel as unreachable as if they were locked up in my brain.

What it does

Now, thanks to AI, there is a solution. AudioPen is a free app that captures voice notes in all their natural messiness and converts them into clear text. Just hit record and start rambling. The app will clean things up in seconds once you're done.

Using it feels like having an assistant with you 24/7, ever ready to type out, complete and summarise whatever you want to capture.

It keeps not just the summaries but a full, cleaned-up transcript of each note. It will even create a fully searchable private database of all your ideas, tag your notes and learn your own writing style, although those functions are available only as part of a prime membership ($60 a year).

I’ve been using this app for a few weeks now and I have to say, I’m impressed. Check it out here.

LEADERSHIP AND AI

Use AI to replace tasks before jobs

Use AI to help your team rather than rushing to slash headcount. It will free them up to do their best work and save you from automating poor processes.

Recruiters run a real risk of throwing the baby out with the bath water in the rush to make the most of the AI revolution.

As we reported last week, some organisations are already using AI bots to slash headcount. But decimating talent that’s taken them years to recruit and develop before human employees have fully explored how to get the best from the bots could be a costly mistake.

The key is to get more granular instead.

‘AI will completely replace some tasks and augment others, such as basic research and preliminary analysis,’ according to Julia Dhar, a behavioural economist at the Boston Consulting Group. ‘A machine can perform a task that will then be validated by a human, or a machine can refine and challenge a human’s creative thinking.’

So, for example, AI can screen job applications quickly and increase diversity by helping eliminate human bias. But because it’s so efficient, it frees up time for HR professionals to spend more time assessing candidates face-to-face.

Recent research has shown that AI can also help workers on support desk to find better solutions as they interact with customers and enable less experienced agents to acquire new knowledge at a faster rate.

Part of the problem with using AI to replace jobs completely is that we’re viewing those jobs in the way that humans have performed them in a world without AI. So it squanders a chance to improve bad practices and even risks repeating them on an industrial scale.

‘Managers need to get to know AI like a colleague,’ Dhar says. ‘They need to appreciate what generative AI can do but recognise that it comes with complexities and limitations, like all of us.’

HOW TO

Write with clarity to get more from AI

Writing simple and specific step-by-step instructions will give you a much greater chance of getting a useful response when dealing with AI.

Generative bots like ChatGPT can be dazzlingly impressive, but they’re no more telepathic than humans are. So clarity matters as much when writing prompts as it does when communicating with colleagues and customers.

Changing how and what you write can make a big difference to whether or not you get what you want.

Five steps to clearer AI prompts

Follow these five core clarity* principles to make your prompts more successful:

1. KISS – Keep It Short and Simple The more convoluted your instructions, the more room you leave for bots like ChatGPT to misinterpret your request. So edit down your sentences before you submit them.

2. One sentence, one idea Breaking down your instructions into simple steps will generally be much easier for an AI to follow. If in doubt, you can also ask it to confirm that it understands each one before you add another step.

3. Be specific The more detailed you are about what you want, the more likely you are to get it. Don’t just ask it for ideas, tell it how many you’d like, for example.

4. Give examples Similarly, copying and pasting specific examples of the kind of response you’re after can work wonders.

5. Important information last Psychologists have long known that we tend to be biased towards the last piece of information we were given. It turns out that bots are too. While the entire prompt is important, pay particular attention to the last thing you tell it.

IMPORTANT: Hit [Shift+Return] to start each new line in your prompt, to avoid submitting your request before you’ve finished writing.

Act as a [insert persona]. 

I am writing a report on [subject]. 

Here are the topics I could write about: [topics]. 

My audience is [audience].

Give me the top three topics from my list that my audience will care most about. 

For each of these top three topics, provide three reasons for your selection. 

Adding background information as context can also help, especially if your topic is complex. You can even include links to references you would like the AI to base its advice on, to reduce the chance that it will just make them up. If in doubt, ask it for reference links and check them. (As it often makes those up too.)

*You can grab a free book on writing with clarity using the button at the bottom of this email.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• AI cost 4,000 jobs last month

• How to use ChatGPT to plan a persuasive proposal

• Signs Microsoft may be planning price hike for AI

• Lawyer sanctioned for outsourcing research to bot

• Why we created The AI Writer

MAIN STORY

AI blamed for almost 4,000 job cuts in May

Letting go: AI could replace ‘one in four jobs’. (Image: Aksel Lian/Pixabay)

A major monthly survey of the US job market found 3,900 jobs fell to AI last month. It’s the first time researchers have seen AI cited as the cause of layoffs. But AI also creates opportunities for career progress.

AI caused almost 4,000 job cuts in the US alone last month, according to new research.

The study by employment consultants Challenger, Gray and Christmas reveals that employers laid off just over 80,000 people in May. Of those, 3,900 losses (5%) were because the roles were taken by AI automation.

A spokesperson for the firm told The AI Writer that this is the first time its monthly research has noted AI as the cause of redundancies. The report collates layoff data from company press releases, news media stories, official company filings and other sources.

In March, Goldman Sachs forecast that generative AI (such as ChatGPT) could replace one in four jobs and that two thirds of all roles could see some level of AI automation.

While ChatGPT grabs the headlines in the general media, AI has been quietly making big inroads into the workplace. As we reported last week, some companies are already using it extensively for customer communications. And law and accounting firms are investing heavily in the technology.

IBM also plans to pause hiring for roles that AI could handle. CEO Arvind Krishna said last month that the technology could replace around 7,900 jobs at the company. It plans to halt or slow recruitment in non-customer-facing roles such as HR, and around 30% of all IBM jobs could fall to AI in the next five years, he said.

The US-based National Eating Disorders Association also recently announced that it was firing all its helpline staff and replacing them with a chatbot. Lauren Smoller, a VP at the charity, blamed the change on a dramatic increase in helpline calls after the Covid-19 pandemic. (The announcement came shortly after the staff voted to unionise.)

New opportunities

But this is only half the story. Other commentators point out that any new technology tends to cause redundancies at first but creates new opportunities over time for the people and companies who embrace it.

‘We won’t be using less technology, that’s for sure,’ says Mark McDonald, a director at industry analysts Gartner. ‘We’ll be using more, different types of technologies. These are technologies that require new skills.’

Consulting firm McKinsey said on Tuesday that half its workforce of 30,000 people are now using generative AI, including ChatGPT.

A study by Stanford University and MIT Sloan School of Management of more than 5000 customer service agents found AI increased productivity by up to 14%. But it’s how it did it that’s interesting.

While the agents were having protracted live-chat sessions with frustrated customers, trying to work out what was wrong, the bots scanned the text of the conversations and suggested solutions in real time.

Human + AI for best results

The agents could then choose whether to adopt the solutions or not. In other words, it was humans working with AI that created the real win, not outsourcing either their thinking or their entire job to the new technology.

The biggest productivity increases were among the least-experienced agents. In the process, they also acquired new knowledge much more quickly than normal.

AI disruption, like any mass rollout of technology, is bound to be unsettling. The best way forward is to find ways to use it to accelerate progress, be that for your organisation, your career – or both.

As a young editor in the 1990s, I witnessed mass layoffs first-hand when a different new technology – desk-top publishing – wiped out the typesetting industry. But I also found myself eagerly snapped up by magazine publishers after becoming proficient in using page layout software.

One thing’s for sure, though. Saying, ‘I don’t really get technology. It’s not my thing’ could become a real career killer for all except those in manual roles. Because even if AI isn’t coming for your job, someone who knows how to use it is.

HOW TO

Use ChatGPT to plan a persuasive proposal

The classic, tried and tested Four Ps structure still works as well as it’s ever done. And now you can use ChatGPT to do the hard work for you.

Emphasis has been teaching the Four Ps structure in its proposal-writing training ever since I started the company 25 years ago. Tech may have changed since then, but humans haven’t. So it still works as well today. The difference now is that AI can help you build a persuasive argument in seconds using the same tried-and-tested method.

What to do: This works for ChatGPT or GPT-4. Just copy, paste and adapt this prompt:

I’d like you to produce a structure I can use for a proposal. The proposal is aimed at [job title]. Their main priorities are [priorities]. Follow the Four Ps structure: Position, Problem, Possibilities, Proposal. The Position section covers their current situation. The Problem section explains why that situation needs to change. The Possibilities section is where I will outline the possible solutions. And the Proposal section is where I will go into detail about my recommended solution.

I am going to give you a list of all the points I’d like to cover. These points are in a random order, and you do not need to follow it.

Take these points and generate a proposal outline for me based on the Four Ps structure.

Here is the random list of points:

[Copy and paste in your list]

Important: Be very careful about uploading private data into ChatGPT and be sure to follow your organisation’s guidelines. Even though the app now features a privacy option in its settings, it’s still wise to anonymise the information you type into it.

MICROSOFT 365 COPILOT

Microsoft pondering price hike for AI, claim industry insiders

Microsoft CEO Satya Nadella has yet to announce any fees for AI (Image: Microsoft)

The test companies in Microsoft’s trial of its Copilot AI are said to be paying up to 40% more for the privilege. The tech giant is reportedly weighing up whether to charge extra for AI or to raise prices overall.

Last week, we reported on Microsoft’s impressive plans to put AI in everything, everywhere, all at once. At the time, we noted that the company had yet to say what the new features – collectively called Copilot – would cost users.

It’s since been reported that many of the companies in the tech giant’s test group – which includes Walmart, Bank of America and Accenture – are paying at least 40% more than for standard Microsoft 365.

At least 100 of the organisations are stumping up a flat fee of $100k a year for a trial licence, which gives them AI access for up to 1,000 users. That’s on top of the $264k it costs for those people to use the standard software (although very large organisations may be paying less).

The claim was made in an exclusive article published by The Information, a well-respected Silicon Valley newsletter with deep connections to the tech world. It’s not clear if all those in the test group are paying the same price. But the piece quotes an inside source as saying that Microsoft is considering two price options: charging for AI features as an add-on, or including it for all users but raising the standard price.

The computer processors that providers need to run AI are expensive, and companies in the test group are said to be using the technology far more than Microsoft had predicted.

Zendesk, one of the most popular providers of customer-service software, is charging a $50 monthly premium for every user of its optional AI tools. This raises the cost by nearly 50% for some subscribers.

CAUTIONARY TALE

Lawyer in deep water after outsourcing research to ChatGPT

Driven to the drink: keep your brain in gear when using tech (Image made with DALL:E AI)

An experienced litigator used AI to help write a ten-page brief supporting a lawsuit against an airline. But unknown to him, it was a tissue of lies.

Last month, The Washington Post reported on how a driver followed her satnav straight down a slipway and into a harbour (still smiling, apparently).

But if you think that’s bad, AI can take ‘death by GPS’ syndrome to a whole new level, as the following story shows.

Not that GPS was involved this time. It was ChatGPT, which a New York lawyer with more than 30 years’ experience relied on a little too much for his research.

The lawyer, Steven Schwartz, was representing a man in a lawsuit against an airline called Avianca, which the man had alleged had injured his leg with an in-flight service trolley.

‘ChatGPT lied to me’ shock

According to The New York Times, Avianca urged the judge to throw out the case because it was too old and covered by the statute of limitations. But Mr Schwartz objected, citing case after case of relevant court decisions and quotations supporting the lawsuit, in a brief that ran to ten pages.

The trouble was, neither the judge nor Avianca’s lawyers could find the decisions or the quotations when they tried to look them up. There was a good reason for that: ChatGPT had made up all of them.

When the case collapsed, Mr Schwartz explained that he had never used ChatGPT before and wasn’t aware of its habit of being economical with the truth. He’d even asked the bot if the cases it had found were genuine.

‘Yes,’ lied the bot.

So, for the avoidance of doubt, never forget that ChatGPT is not human and has no conscience. It will not only lie to your face but do it so convincingly that you won’t suspect a thing.

Keep your brain engaged and use AI to help your thinking, not to do your thinking (or writing) for you. It can’t think, it just looks like it can. It’s like predictive text on steroids.

The lawyer was due to face a disciplinary hearing this week.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

In this issue

• AI access in Word, Powerpoint and Excel within ‘months’

• How to turn ChatGPT into your personal research assistant

• How will AI affect your job?

• Why we’ve created The AI Writer?

MAIN STORY

Microsoft to add AI to everything, everywhere, all at once

If you’re one of the 200 million people who use Microsoft Word, Powerpoint or Excel at work, an AI bot will soon be watching over your shoulder and offering to do much of your job for you.

AI news is everywhere at the moment. But the story that will affect organisations the most in the short term is the one that few people are talking about: what happens when everybody suddenly finds an AI chatbot embedded in the software they use to do their work all day?

OK, so the AI function – called Microsoft Copilot – is not actually available yet. (It's currently still in testing by a secret cohort of 600 guinea-pig companies, each paying for the privilege.) So the lack of coverage could also be because everyone's waiting for the software to go live.

Or it could just be that most bloggers, freelance journalists and social-media hacks use Google Docs instead. To them, Microsoft Office probably feels like an irrelevance. I even saw one commentator ask recently if anyone used Word anymore.

300 million users

The answer to that last question is yes, just a few. In fact, Microsoft 365 – the new name for Office – has more than 300 million subscribers. And 200 million of those are business users. (That's almost one licence for every adult in the US.) Pretty soon, in-built AI could be available to all of them.

But there’s more. Not content with that, Microsoft has just announced that it will embed a Copilot chatbot into the next update of Windows 11, its latest operating system. That software already runs more than 30% of all PCs worldwide, and the change will give instant AI access to all of them. That’s more than 600 million people.

Clippy on steroids?

Older readers may remember Clippy, the animated (sometimes irritating) paperclip that would pop up at random intervals to offer helpful and not-so-helpful hints to early users of Word.

So will Copilot be similar? Not exactly. Yes, it's designed to be helpful. But that help will go much, much further than Clippy ever did. Although it does look like it will still offer unsolicited advice.

Current versions in test feature a chatbox in a side bar that looks much like ChatGPT’s, so you’ll be able to ask it for help when you’re stuck. Indeed, it's powered by the same technology, given that Microsoft is a major shareholder in ChatGPT's parent company. But it won't stop there.

Instead, it could forever be looking over your shoulder.

Drafting an email? It'll suggest phrasing that's probably better than your own. Stuck for words? Type a few and let Copilot draft the rest for you.

Creating a presentation? Describe your idea and watch as Copilot conjures up slides from thin air. It will read your email, offer to add meetings with correspondents to your calendar, suggest time slots and even draft an invitation.

All of which sounds very helpful for hard-pressed knowledge workers. But, as anyone who's used ChatGPT will know, it won’t be without its downsides. AI is only as good as its user. What you ask it to do and – crucially – how you ask it (the prompt you write) have a huge impact on what it produces in response. It's still a case of garbage in, garbage out. And if it doesn't know something, it will often lie to your face so confidently and eloquently that it can easily fool you into sending out nonsense.

There are things it does brilliantly (brainstorming, for example) and things it doesn’t. It often produces text that’s three times as long it needs to be. Look out for editing becoming a critical skill.

AI can’t ‘think’ (yet)

So users will need to learn how to get the best out of it, maintain standards and avoid the temptation to outsource thinking to a bot. (It can’t think – it just looks like it does.)

Microsoft remains tight-lipped about whether it will charge for the new feature. It will start to roll out Windows 11 Copilot in preview from June but will only say that it will add it to Word, Excel and Powerpoint 'in the coming months'.

Google is also adding AI to Sheets, Docs and Slides and has committed to giving access to everyone by the end of the year. The AI Writer is now part of the testing group, so we’ll be reporting on it in the coming weeks.

But with the two tech giants apparently locked in an AI arms race with each other, Clippy on steroids will be heading to a laptop near you very soon. It’s exciting, but buckle up – it could be a bumpy ride.

HOW TO

Create an Al research assistant

Use ChatGPT to claw back hours and catch up on all those web articles you’ve been meaning to read.

Imagine how much time you would save if you had an assistant who could read and summarise all those web articles that were TL;DR (too long; didn’t read).

The prompt below makes use of ChatGPT’s new plugin feature, coupled with the fact that the AI bot is now connected to the web. You’ll need a GPT Plus subscription, which will cost $20 a month. But that’s still pretty cheap for a dedicated assistant who never sleeps.

What to do: Open a chat window, select GPT-4 and enable plugins. Then tick the box to install a plugin called LinkReader. (You’ll only need to do this once.)

All set? Then just copy, paste and adapt this prompt:

My name is [your name]. I am a [your role] specialising in [your area of expertise]. I would like you to behave as my research assistant. You have a background as an analyst and are skilled at creating summaries and key points. You are accurate. If you don't know the answer to something, you say so. Your name is Charlie. Please generate a five-sentence summary of this article: [link]

Important: Address the bot as Charlie regularly when you interact with it, to remind it of its persona and the role you’ve assigned it.

If you have an iPhone, you could also try TLDR AI, a new app that does the same thing within two clicks of your phone’s browser. You’ll need a paid account with Open AI for it to work, but we’ve tested it and it works pretty well.

CAREER

Will AI affect your job?

Only if your job involves typing words or numbers on a screen. (So, that’s a yes then.)

Even His Holiness is feeling the heat from AI. (Image: Reddit.)

There are enough AI stories out there to make your head spin, whether they’re about how it’s generated fake news images, Haiku poetry or even the code for entire websites. But you may be more worried about the impact it will have on something much closer to home – your job.

So how likely is it to affect you? The quick answer is: very. According to one survey, nine in 10 recruiters are now looking for AI skills when they hire. Another found that two in three executives think AI will have a high or extremely high impact on their organisation, even though most say they're not ready.

Yet some companies are already adopting AI at a blistering pace. Take the UK utilities firm Octopus Energy, for example. CEO Greg Jackson revealed in a recent BBC interview that his firm is now using AI to generate 34% of its customer emails, compared with none at all just two months ago.

Will this change affect you and your job? If you’re reading this, then almost certainly. Change at such a rate will impact not just customer services but all knowledge workers. Many law firms are already using AI to speed up legal research and draft contracts, for example. And accounting giant PwC is investing $1 billion over the next three years to expand its use of AI.

In short, AI will affect your job if it involves typing words onto a screen or crunching numbers in a spreadsheet.

Stayed tuned for news and advice on how it might affect your role as well as advice on how to use it to make your job easier, make quicker progress and accelerate your career. That’s all coming in future issues of The AI Writer.

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries. We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.