Thought #20 - Skills, Scrutiny and Sacking

Training drives, regulatory pushback, and another round of layoffs

Hi lovely humans,

This week’s AI stories are all about what people are learning, where the pressure’s building, and who’s being let go. We’ve seen new training initiatives for teachers, civil servants and local communities - alongside a few big layoffs, poaching sprees, and a steady stream of “what does this mean for jobs?” reflections. It’s a skills week, with a good dose of scrutiny.

Our Week in AI

Actually Using Our Platform for Learning

I’ve been experimenting with using our chatbot Dotly for more hands-on learning - and Data Protection has been first. I enjoyed going through my own understanding of the principles and how they relate to my role. Much better than reading boring pages and clicking through a multiple choice exam.

It’s early days, but it feels like a more useful way to check what you know (and don’t). Next, we’re going to look at some AI principles - bias and explainability.

Dotly teaches the GDPR principles

Understanding Long Contracts

While I’m not saying AI should replace lawyers, I am currently negotiating a big contract (for us) and have found ChatGPT to be really helpful.

Prompt - We’re the seller in this contract - can you go through call out any issues or anything I need to be aware of (in terms of responsibilities)?

This highlighted some things I need to consider in terms of data protection, reporting, and some of the more subtle bits about delivery timelines and liabilities.

Again, not a replacement for lawyers or any professionals, but in a situation where I would just be reading 60 pages myself - really helpful.

AI New Releases

It’s all about Anthropic this week - transparency, examples and training.

Powered by Claude
A new showcase from Anthropic showing how real companies are using Claude - legal, customer service, mental health, that sort of thing.

Transparency framework from Anthropic
They’ve also published a proposal for more transparency around model safety - with a first step focused on visibility rather than regulation.

Claude courses for builders (tech not construction)
A new set of short online courses for people building with Claude. Basic, but could be useful for technical teams getting started.

AI News

Microsoft, OpenAI and Anthropic fund teacher training
AI companies are paying for teachers in New York to get trained in how to use ChatGPT and other tools in the classroom. Still early, but could be a model for others.

NHS hospitals using AI to speed up MRIs
Hospitals in Hull say AI is saving time and reducing staff stress. One of the more practical use cases we’ve seen.

UK government teams up with Google
A new deal to save money on old tech and train 100,000 civil servants in AI. Light on detail for now, but part of a bigger trend.

Neighbourly, Microsoft and Onside launch local AI skills drive
A new campaign to support AI learning in cities like Manchester. Focused on young people and communities — not just schools.

AI “fingerprints” found in medical papers
New research shows signs of AI-generated content in scientific publishing - not always malicious, but sometimes misleading.

Canva gives staff a week to play with AI
Canva’s giving 5,000 people a full week to learn and experiment with AI, no strings attached. Might spark ideas for internal L&D.

Meta poaches Apple’s AI talent
Meta’s continuing its hiring spree, luring top AI engineers from Apple. Still very much a talent land grab.

$20M raise for AI in construction procurement
Parspec’s software helps construction firms with bidding and procurement. One of the more specific (and less hyped) use cases.

Adobe expects AI to boost Prime Day sales
Their forecast? £23bn in online spend, with AI-powered personalisation and advertising playing a growing role.

Grok deletes posts praising Hitler
X’s AI chatbot posted (and then deleted) content supporting Hitler. A stark reminder that moderation’s still very much an open problem.

AI scams and security risks
A fake voice of a US official was used to call foreign ministers. Real-world consequences, with no easy fixes yet.

Terrorist groups experimenting with AI
Mostly content generation and planning. Not new, but governments are starting to talk more openly about it.

European AI CEOs urge rethink of EU AI Act
Leaders from Mistral, Sana and others are asking for a pause on the EU AI Act - they say it could slow down innovation, especially for open source.

Microsoft cuts 9,000 jobs
Layoffs and game studio closures as part of a wider reorganisation. A reminder that AI investment doesn’t always mean job creation.

Layoffs slightly down year-on-year
Despite the doom and gloom, overall US job cuts are 2% lower than this time last year.

Not Quite News, But Worth a Read (or Listen or Watch)

  • AI and the grad job market
    The Guardian warns that AI risks squeezing out younger workers. Not new - but the tone is shifting from concern to policy.

  • All jobs gone in 20 years?
    A futurist says AI will replace all human labour by 2045. Probably not - but the economic shift is worth paying attention to.

  • Paid to fix AI’s mistakes
    A good explainer on the hidden jobs behind AI - people checking facts, fixing tone, filling in gaps. Not glamorous, but growing.

  • Will AI change how we do maths?
    Some researchers think AI could help us discover new maths, not just solve old problems. Still speculative, but interesting.

LinkedIn AI Poll

Last week I asked if people knew what Agentic AI is - apparently most people do (I found this surprising as I wouldn’t confidently claim this…)

Only one person was brave enough to give a definition in the replies, and I think it’s quite good “it's a system or process that performs tasks independently. It can make decisions and does not need human intervention to complete tasks or initiate processes or even make decisions within the framework in which it works or for the purpose its been built.”

Vote in this week’s poll - please!

This week we’re focusing on data privacy - specifically how you feel about OpenAI using what you share in ChatGPT.

Our Blogs and Resources

Our Learning and Research Manager Carys has been exploring ChatGPT helping with the pain of Google Sheets formulas:

Final Thoughts

As always we hope this was helpful!

Feel free to share this with anyone who might find it useful.

Next week, [UPDATE]

Laura
Always learning