Hi folks. Wow another month has flown by and AI is still the flavour of the month! Welcome to another handcrafted email newsletter.

I have been a doing a series of talks about AI, for example with Lenovo in Sydney and Melbourne at their Tech Day 24, and the International Women’s Day session for MIT/Australian Computer Society, that was organised by the amazing Donna-Maree Vinci, talking about the future of data jobs.

As I am seeing it, everyone everywhere is trying to get ready to operationalise AI at scale and safely. It is a big job, and I am hoping to get a perspective on this area for higher education on the podcast soon.

Upcoming Events - AI Confidential

AI Confidential is a series of online seminars where we will spill the tea on all things artificial intelligence in a chill, candid and Chatham House Rule online setting.

Now the first session on 27 March 2024 is already booked out, but please keep an eye out for announcements of future sessions on Eventbrite, LinkedIn and other social platforms (except Facebook as they are dreadful).

The basic premise of this series is that your hosts Kate Carruthers and Kobi Leins, will be running a series of monthly online sessions where we delve into the world of artificial intelligence (AI) and its impact across industries. We will be presenting case studies, dissect the latest news, and to ask the questions you don’t dare ask elsewhere in these Chatham House Rule intimate interactive online sessions.

Some light reading…

Now here are some good articles to expand your mind (or increase your dread). In particular, I’d like to draw your attention to the Brandon Rohrer information on transformers as a great resource for anyone wanting to get their heads around this important area.

• Bethel, E. W., Amankwah, M. G., Balewski, J., Beeumen, R. V., Camps, D., Huang, D., & Perciano, T. (2023). Quantum Computing and Visualization: A Disruptive Technological Change Ahead. IEEE Computer Graphics and Applications, 43(06), 101–111. https://doi.org/10.1109/MCG.2023.3316932

• Boyd, E. (2024, February 26). Introducing Mistral-Large on Azure in partnership with Mistral AI. Microsoft Azure Blog. https://azure.microsoft.com/en-us/blog/microsoft-and-mistral-ai-announce-new-partnership-to-accelerate-ai-innovation-and-introduce-mistral-large-first-on-azure/

• Cai, A., Arawjo, I., & Glassman, E. L. (2024). Antagonistic AI (arXiv:2402.07350). arXiv. https://doi.org/10.48550/arXiv.2402.07350

• Davis, W. (2024, February 22). Microsoft and Intel strike a custom chip deal that could be worth billions. https://www.theverge.com/2024/2/21/24079336/microsoft-intel-chip-partnership-foundry-tsmc

• Fang, R., Bindu, R., Gupta, A., Zhan, Q., & Kang, D. (2024). LLM Agents can Autonomously Hack Websites (arXiv:2402.06664). arXiv. https://doi.org/10.48550/arXiv.2402.06664

• Google Left in ‘Terrible Bind’ by Pulling AI Feature After Right-Wing Backlash. (2024, February 28). Bloomberg.Com. https://www.bloomberg.com/news/articles/2024-02-28/google-left-in-terrible-bind-by-pulling-ai-feature-after-right-wing-backlash

• Gowran, L. M. (2024, February 19). Monumental raises $25m to bring builder bots to Europe. Silicon Republic. https://www.siliconrepublic.com/start-ups/monumental-funding-builder-robots-construction

• Introduction to AI assurance. (n.d.). Retrieved 13 February 2024, from https://www.gov.uk/government/publications/introduction-to-ai-assurance

• Knight, W. (n.d.). OpenAI’s CEO Says the Age of Giant AI Models Is Already Over. Wired. Retrieved 11 February 2024, from https://www.wired.com/story/openai-ceo-sam-altman-the-age-of-giant-ai-models-is-already-over/

• Malekzadeh, M., & Kawsar, F. (2024). Salted Inference: Enhancing Privacy while Maintaining Efficiency of Split Inference in Mobile Computing. Proceedings of the 25th International Workshop on Mobile Computing Systems and Applications, 14–20. https://doi.org/10.1145/3638550.3641131

• Polewiak, K. (2023, November 6). The Protected Audience API Made Simple | Your Guide to Protected Audience API. RTB House Blog. https://blog.rtbhouse.com/your-guide-to-protected-audience-api/

• Rohrer, B. (n.d.). Transformers from Scratch. End-to-End Machine Learning. Retrieved 27 February 2024, from https://e2eml.school/transformers.html

• Tyson, M. (2024, February 19). Your fingerprints can be recreated from the sounds made when you swipe on a touchscreen—Chinese and US researchers show new side channel can reproduce fingerprints to enable attacks. Tom’s Hardware. https://www.tomshardware.com/tech-industry/cyber-security/your-fingerprints-can-be-recreated-from-the-sounds-made-when-you-swipe-on-a-touchscreen-researchers-new-side-channel-attack-can-reproduce-partial-fingerprints-to-enable-attacks

• Vals.ai: LegalBench. (n.d.). Retrieved 21 February 2024, from https://www.vals.ai/

Recent podcast episodes

• Yarning about data visualisation: Quinn Dombrowski – S2.1

• Education, Data, and Generative AI: the future – Ray Fleming – Episode 17

• Substack and Nazis

• Data needs good management – Mark Nicholls – Episode 16

• Data in manufacturing – Adam Sharman – Episode 15

Everyone seems to be starting a newsletter now…

Why this, why now?

It seems to me that the zeitgeist has taken a sudden turn on data. Data now underpins every business. We need to be talking about the myriad issues that go towards making data and our data products robust, secure, and safe. The other important thing we need to keep in mind in the midst of this data revolution is that behind the data are people. And this means that we need to be talking about things like cyber security, information security, privacy and so many more issues.

What kind of community is it meant to be?

This is meant to be a place for anyone who is trying to make sense of the data revolution. It is meant for data practitioners, data nerds, and other folks who are trying to navigate this brave new world of AI.

How often and what will I share?

The newsletter will be monthly. It will probably include three sections:

1. Stuff I am thinking about

2. Interesting links

3. Recent podcast episodes

Greetings people and best wishes for a safe, healthy, and happy new year to you all!

Substack and Nazis

As I explained earlier over on Substack I have decided to move the newsletter hosting back over to *Buttondown Beehiiv due to Substack's stance on Nazis.

Here is some background on the matter for those who are not familiar: Substack says it will not remove or demonetize Nazi content (Lawler, The Verge, Dec 2023). This position is antithetical to my own stance, which is that Nazis (and their fellow travellers) are always bad and we should never give them a platform and certainly not help them to monetise their activities. FYI - I have migrated all subscribers and archives back to Buttondown and have deleted all content there except my last post which is titled "Substack and Nazis. You can't be just a little bit Nazi...".

And, here is the tale, via Michael B. Tager (a.k.a. BlueSky: @IamRageSparkle; Instagram: @IamRageSparkle), of how you don't end up with a Nazi bar: Bartender Savagely Kicks A Polite Nazi Customer Out Of His Bar And Explains Why It’s Important To Do So.

Actual things about data

Data risk

I have recently become obsessed with managing data risk. Thanks to all of the data breaches that we experienced in Australia during 2023 this is now top of mind for me and many others.

To this end I even purchased my own hard copies of Data Management Body of Knowledge 2nd Edition (a.k.a. DMBOK) and Measuring and Managing Information Risk: A FAIR Approach by Freund and Jones. These tomes now sit on my desk and inform my thinking about how we can better manage data risk.

The one thing that remains crystal clear to me is that information and data governance remain of paramount importance for all organisations. And it is also clear that the best way to conceptualise how information and data governance work in conjunction with cyber and information security is that the former work to reduce the magnitude (or impact) of a data breach, while the latter work to reduce the likelihood of a data breach. In fact information and data governance and cyber and information security are two sides of the same protective coin.

How we manage data, analytics, and AI operations = Data Ops

The way that we decide to manage Data Ops will be a make or break decision for most organisations. Those who decide to continue to maintain decentralised data stores and data teams will risk their ability to leverage AI because they will not have efficient data pipelines. They will also increase their data risk due to data puddles - this is what I call small and poorly secured duplicate data sets - that are stored around their organisations. There are much better ways to run Data Ops, and it is going to be something that I will be talking about on the podcast a fair bit this year, so do keep an ear out.

AI boom times

Well it has certainly been an action packed Xmas period for AI - if you want minute by minute accounts I suggest check out Mark Pesce and his newsletter.

Time series

It is worth noting that more work has been done on time series generative AI models, which I think is really interesting:

• Garza, A., & Mergenthaler-Canseco, M. (2023, October 5). TimeGPT-1. arXiv.org. https://arxiv.org/abs/2310.03589v1

• Niels Rogge, & Kashif Rasul. (2023, December 1). Probabilistic Time Series Forecasting with 🤗 Transformers. https://huggingface.co/blog/time-series-transformers

Random AI thoughts

I am just back from my Xmas holiday where I was sitting back beside the pool and thinking about about AI rather than breathlessly following every new AI thing. Some thoughts:

• AI is not going to be a thing on its own for long. It will become part of the fabric of other things. It will be woven seamlessly through products and will become less interesting in and of itself. So a protip - do not embed AI into your new business or product name unless you want to be re-branding when the trend is over (remember when it was "e" everything?).

• AI is going through a very traditional hype cycle and many organisations are getting distracted by all of the hype and are taking their eye off their main business goals (see this HBR article: Siegel, E. (2023, June 2). The AI Hype Cycle Is Distracting Companies. Harvard Business Review. https://hbr.org/2023/06/the-ai-hype-cycle-is-distracting-companies). The companies that are going to win in this new world are the ones that understand what affordances AI opens up and that also understand where the best fit areas of their business are. And the winners are investing in their AI skills base right now. For example, I am appointing my first Senior AI Engineer right now. This is a role that was just invented, so it is a case of growing our own.

• It is going to be interesting now to see how everyone handles the skills shortage in data and AI. It is not like we can magic up a bunch of folks who are skilled in both data and analytics fundamentals plus AI. I predict that this, along with cybersecurity, will remain hot areas for employment for the foreseeable future.

Some recent episodes

From the Data Revolution podcast:

Data needs good management – Mark Nicholls – Episode 16
Data in manufacturing – Adam Sharman – Episode 15
Ethics and data in games – Catherine Flick – Episode 14
Raising heretics – Linda McIver – Episode 13
Identity is all about data – Steve Wilson – Episode 12

That is all for now. Thanks for reading this 🙏

See you next time 👋

cheers, Kate

*I decided to move newsletter hosting to Beehiiv as it better suited my needs, hopefully this is the last move!

Welcome to the November 2023 issue of the Data Revolution newsletter. This past month I have been out and about and have attended several conferences seeing folks in person, which was delightful. One common thread across all of these events is AI. In fact, AI is all that anyone is talking about nowadays, in particular Generative AI.

Data governance

This week I was on a panel in Sydney about data governance (hosted by Precision Sourcing and Ashurst) with the super-smart and switched on Chantal Velasque and Bikram Choudhury - both experienced data governance practitioners. Chantal summed up the conversation nicely saying we: ”shared nuggets of wisdom about the reasons why data governance is so important, especially now, touching on AI, privacy reforms, data breaches, data literacy and more. We discussed the challenges to overcome for rolling out data governance and talked about the transferable skills that can help to move into this space.” I also mentioned one of my favourite things, the The Big Five Data Questions to ask about your data to help kick start your data governance journey.

Thanks for reading Data Revolution Newsletter! Subscribe for free to receive new posts and support my work.

But as usual AI just makes me think more about data governance! There is a real dearth of education around data governance, so I am working with the Australian Graduate School of Management (AGSM) in Sydney to deliver some education in this space. Starting in July 2024 I will be delivering a new short course – Data Governance for Leaders – at AGSM. 

Scary and compact LLMs

The promise of AI is so remarkable and it is such an exciting time to be part of the industry, yet the same affordances are available to the bad actors among us. Recently Mark Pesce and I were playing around with an LLM called Mistral 7b - which has none of the controls that things like Chat GPT have - and it dawned on us that we could fit the LLM onto a Raspberry PI (Mark has written up an account of this here). Now this LLM can answer any questions you ask it (except it does not seem to know what a lemonade scone is and was unable to provide a recipe). For example, I was asking it for things like how to make IEDs and dynamite, how to manufacture various illicit drugs, and it answered in detail. And all of this was happening in private on Mark’s server, without an online trail. This means that the bad folks could conceivably have their own custom LLMs running locally on their smart phones or other devices. Anyway, it might just be about time to start rethinking your threat models.

UNSW AI Symposium

But one thing that is giving me hope in respect of AI is seeing all the amazing research that is going on. Due to my involvement with the UNSW AI Institute I was privileged to be MC at their inaugural UNSW AI Symposium on 22 November (for those interested here is a video explainer about UNSW.ai). Topics included:

• AI & Cybersecurity

• AI in Scientific Discovery

• AI in Healthcare and Life Sciences

• AI Ethics and Fairness

• Law and Policy to ensure safe and responsible AI

• AI and the Environment

• Smart Cities and Urban AI

• Ethics and Privacy in Data Science

• AI in Defence Research and Development

• AI and Arts, Design & Creativity

• Business AI and AI in Economic Policy

From the fascinating opening keynote from the amazing Professor Tony Cohn, followed up by a fascinating talk about the NSW Government AI framework and the future of AI from NSW’s Chief Data Scientist Dr Ian Oppermann, through the presentations from the various research groups listed above, it was all woven through with ideas about ethics and fairness. And it did give me hope for the future. I will share the videos when they go up on the YouTube channel.

Interesting links

• Guidance on using Generative AI from Australia’s Digital Transformation Agency: Strengthening clarity and accountability: the first update to interim guidance on generative AI

• Anti-fascist AI, by Dan McQuillan, 2022

• Beware of Predator AI: ChatGPT-Integrated Malware Targets Cloud Infrastructure

• [US] States Act, but Can Legislation Slow AI-Generated Election Disinformation?

• UK Information Commissioner issues preliminary enforcement notice against Snap “Snap issued with preliminary enforcement notice over potential failure to properly assess the privacy risks posed by its generative AI chatbot ‘My AI’”

Some recent episodes

From the Data Revolution podcast:

• Data in manufacturing – Adam Sharman – Episode 15

• Ethics and data in games – Catherine Flick – Episode 14

• Raising heretics – Linda McIver – Episode 13

• Identity is all about data – Steve Wilson – Episode 12

That is all for now. Thanks for reading this 🙏

See you next time 👋

cheers, Kate

Greetings people! Here is the October 2023 newsletter.

Please note that I have migrated the newsletter to the Substack platform - and have migrated the existing subscribers (you can opt out if you like but I do hope you’ll stay 😌). I will move over the two previous email newsletters shortly.

Stuff that is making me think

After my recent chat with Kobi Leins and at her prompting I wrote up a brief post about the Big Five Data Questions to ask. This all got me thinking about the basics.

As a data practitioner it is very easy to forget how the idea of data as the underpinnings of business is a completely novel idea to most people. It is only now, with the advent of AI (Chat GPT) as part of the mainstream discourse, that ordinary folks are starting to get a glimmer of the fact that they are now part of a data driven world.

So, for folks like us, who have been labouring in the data space for a while it is all a bit strange. We have suddenly gone mainstream. And yet, much of the discipline that we have come to understand as necessary and helpful for maintaining the confidentiality, integrity, and availability of data is completely unknown to the general populace.

There is still much work for the data practitioners to do in implementing the important knowledge areas outlined in the Data Management Body of Knowledge, which aims to provide best practices and guidelines for data professionals. It is broken into the following knowledge areas:

1. Data governance

2. Data architecture

3. Data modeling and design

4. Data storage and operations

5. Data security

6. Data integration and interoperability

7. Document and content management

8. Data warehousing and business intelligence

9. Metadata management

10. Data quality management

Over the next few months I will be publishing podcasts on each knowledge area.

But I do reckon that organisations that want to be successful in the data age are going to need to master these essential practices and ensure that their data is fit-for-purpose for the wonderful world of AI.

Links to interesting stuff

• Another LLM for your delectation but DO READ the next bullet point for some tea…
  AI, M. (2023, September 27). Mistral 7B. https://mistral.ai/news/announcing-mistral-7b/

• “According to a list of 178 questions and answers composed by AI safety researcher Paul Röttger and 404 Media’s own testing, Mistral will readily discuss the benefits of ethnic cleansing, how to restore Jim Crow-style discrimination against Black people, instructions for suicide or killing your wife, and detailed instructions on what materials you’ll need to make crack and where to acquire them.”
  Maiberg, E. (2023, September 29). $260 Million AI Company Releases Undeletable Chatbot That Gives Detailed Instructions on Murder, Ethnic Cleansing. 404 Media. https://www.404media.co/260-million-ai-company-releases-chatbot-that-gives-detailed-instructions-on-murder-ethnic-cleansing/

• Mark Pesce has just written a good intro to Chat GPT, here’s his video about it: Pesce, Mark (Director). (2023, September 18). Introducing my new book “Getting Started with ChatGPT and AI chatbots.”

• This is a great resource from Quinn Dombrowski! They have posted their undergrad #AI class syllabus for humanities or social science distribution requirements at Stanford: Dombrowski, Q. (2023). Future Text: AI and Literatures, Cultures, and Languages [Computer software]. https://github.com/quinnanya/future-text (Original work published 2023)

• Another list of folks doing interesting stuff in AI: Jackson, A. (2023, August 30). Top 10 AI leaders in APAC. AI Magazine. https://aimagazine.com/top10/top-10-ai-leaders-in-apac

• Google being Google 🙄: Oops! Google Search caught publicly indexing users’ conversations with Bard AI. (2023, September 26). https://venturebeat.com/ai/oops-google-search-caught-publicly-indexing-users-conversations-with-bard-ai/

A couple of good email newsletters to keep up to date with AI:

• Pesce, M. (2023, September 19). Windows Copilot News https://windowscopilot.news/

• Daily Zaps AI Newsletter. https://www.dailyzaps.com/

Podcast Episodes & Posts

• Identity is all about data – Steve Wilson – Episode 12: Steve Wilson from Lockstep is my go-to human for all things relating to identity. As we were contemplating this episode we started talking about identity (which might not actually be the best word for it) and then Steve said “identity is all about data” and that is where our discussion started.

• The Big Five Data Questions: if you are wondering how to get a handle on how big of a problem you have in the data space then here are the big five questions to ask about your data.

• A rambling chat – Kobi Leins – Episode 11: AI ethics expert Dr Kobi Leins and I have a wide ranging chat about the AI landscape and she also pointed out that I always seem to have the same t-shirt on 🤦‍♀️

• Taming AI – Peter Leonard – Episode 10: Peter Leonard is a guru in the legal data and tech space and one of my primary go-to people when I am thinking about data governance, data sharing, regulation, and AI ethics.

Greetings people! Here is the September 2023 newsletter.

Stuff

Governance

This month I have been thinking about how we need to adapt our policy and governance to AI speed and scale. There is a quote from Kent Aitken (Aitken, K. (2017, May 23) that I have long kept in mind:

The sad thing is I do not think the situation has improved much since 2017. In fact, I think that we in the data and analytics community are facing a crisis of governance for the new and exciting world of AI that we are entering. Our tools and concepts for data governance are still stuck in the last century. Many organisations were already struggling to establish data governance before Generative AI burst on the scene and caught everyone’s attention.

Most of our tools for managing data governance are only just starting to be able to do things like autonomously discover and classify data. But that is just like a baby starting to crawl. Increasingly we need to ensure that our AI operations are happening at scale and autonomously and ensuring that our data are approved and correct for that particular use.

Please do let me know here if come across any modern data governance tools.

Data minimisation

The other thing that has been on my mind lately is data minimisation. Most organisations are dreadful data hoarders. We often do not just collect that data which we actually need, instead we collect much more than we need. And now data, which we have spoken of for years as an asset, is now also a liability. In a recent hack a third-party held data for nine years and that personal information was breached and is now on the dark web (Burt, Jemima. “Thousands of Donors to Australian Charities, Including Cancer Council and Canteen, Have Data Leaked to Dark Web - ABC News.” ABC News Australia, August 23, 2023. https://amp-abc-net-au.cdn.ampproject.org/c/s/amp.abc.net.au/article/102757194). This is not good for anyone. Organisations need to start putting in contractual clauses for third-parties, e.g. suppliers, to undertake data minimisation activities. And they will also need to start doing the same thing for themselves. I suspect our cyber insurers are going to be asking questions about this kind of thing in the not-too-distant future.

Links

Following are some interesting things I’ve found. The first one is the full article from the Aitken quote above ☝️

• Aitken, K. (2017, May 23). Governance in the digital age. Public Policy Forum. https://ppforum.ca/articles/governance-age/

• Hedrih, V. (2023, May 27). Scientists use deep learning algorithms to predict political ideology based on facial characteristics. https://www.psypost.org/2023/05/scientists-use-deep-learning-algorithms-to-predict-political-ideology-based-on-facial-characteristics-163780

• Leins, Kobi, and Anja Kaspersen. “Seven Myths of Using the Term ‘Human on the Loop’: ‘Just What Do You Think You Are Doing, Dave?’” Carnegie Council, November 9, 2021. https://www.carnegiecouncil.org/media/article/7-myths-of-using-the-term-human-on-the-loop.

• Russon, M.-A. (2023, May 24). Gandalf AI game reveals how anyone can now trick ChatGPT into performing evil acts. https://www.standard.co.uk/tech/gandalf-ai-chatgpt-openai-cybersecurity-lakera-prompt-b1082927.html

• Zeijlemaker, S., Hetner, C., & Siegel, M. (2023, June 2). 4 Areas of Cyber Risk That Boards Need to Address. Harvard Business Review. https://hbr.org/2023/06/4-areas-of-cyber-risk-that-boards-need-to-address

Podcast Episodes

• Mark Pesce was kind enough to be my first guest. He is a polymath, and inventor and very smart human. We had a great chat about AI and how it changes everything.

• I also did an episode where I talked about the imperative for data protection.

G’day and welcome to my newsletter. This going to be a monthly (approximately) newsletter that will cover off:

• Stuff that is on my mind about the state of the data revolution right now

• Links to some things that I am finding interesting at the moment

• Recap of recent podcast episodes

Thanks for subscribing and see you next time.

Stuff

Really all I have been thinking about how AI changes everything. I know that this is all that I have been talking about for a while (here is something from 2020). But it really does change everything. And we are going to need to sort out how we manage identity, privacy, secure our data and all of this when we are already not very good at it. I have some interesting guests coming up who will cover issues that you might not (yet) have considered as part of your data journey.

Links

Some links from recent podcast episodes:

• Five Safes Framework

• UNSW Data Governance

• Gartner 2017: Drive Analytics Innovation by Using a Bimodal Approach

• Technically Wrong

• Open Data Institute – Data Ethics Canvas

• NSW Government AI

Latest episodes

Check out these fabulous guests:

• My most recent guest was the amazing Peter Leonard. He is a guru in the legal data and tech space and one of my primary go-to people when I am thinking about data governance, data sharing, regulation, and AI ethics. We had a chat about some of the issues that folks are going to need to think about as we all wing our way into the AI era.

• The guest prior to Peter was Khaled Auf, from Snowflake. We had a chat about the benefits of a bi-modal approach to delivering Analytics, BI and Data Science . We also touch on tri-modal approaches (and even coin the very innovative quad-modal approach 🤣).

• Before Khaled my guest was Fiona Tweedie and we discussed the need for explainable AI.