Thought #8 - From AI Image Rules to Mindmaps and Hiring Debates

Hi lovely humans,

We’re just going to send a short update this week - as always things are busy in the world of AI, but we’ve been super busy this week!

What We’ve Been Up To

Some happy news (and what we’ve been busy with), Taught by Humans won £15k in a pitch competition at University of Bristol - read more about us and the other teams who won.

We’ll be putting this towards a specific project to help upskill people into entry level AI jobs - more details to follow soon!

AI New Releases

OpenAI have added some rules about image uploads

They have clarified something I was wondering - “You must have permission to edit images you upload, including consent from any people in them.”

Google Gemini now has DeepResearch

Very similar to OpenAI’s DeepResearch and Perplexity. One benefit being the Gemini models having better access to live information from the internet (and detailed Google Scholar information).

Moody Monday from OpenAI

We’re pretty sure this is an experiment about AI personalities. If you’d like to chat to a sassy AI via a CustomGPT, then this is for you.

Google NotebookLM Mindmaps

Not necessarily a new feature this week, but we’ve only just found out about it. We’re big fans of NotebookLM from Google - we’ve used it to turn recordings of me talking about the product into grant applications and website content. But what we hadn’t seen before was the MindMap options - which we found useful for visualising content for our pitch deck (but can see useful applications in studying, product design, content creation). Give it a try.

ChatGPT Memory

Not quite a new release, but a weird ominous tweet from Sam Altman indicating OpenAI might be changing how memories work in ChatGPT. We’ll be on the look out for the press release with more details.

AI News

Some polarising news stories this week on public sector use of AI and companies AI hiring policies:

• The Ministry of Justice has been called out by State Watch for creating an alleged “murder prediction” algorithm with concerns around bias in the model and the data being used (read more in the Guardian coverage)

• Shopify has announced managers must prove new hires roles cannot be done by AI - in a leaked memo from the CEO which has received very mixed comments

• The EU Commission has launched an action plan to make the EU more competitive when it comes to AI - including 200 billion euros for AI development and 13 AI factories

Also worth a read

Not exactly news but we thought these were useful and worth reading:

• Stanford AI Index 2025 - which notes AI optimism rising (but big differences country to country) and AI becoming more embedded in everyday life

• Miro (the collaborative online whiteboard company) found some unexpected differences in AI attitudes between millennials and GenZ

LinkedIn AI Poll

Last week’s poll on whether OpenAI’s image generation model creating in particular styles, eg Studio Ghibli, had over 50% voting definitely yes!

Interesting, we’ve seen a lot of different people voting on this poll than previous weeks.

Vote in this week’s poll - please!

It’s about the regulation vs innovation debate

Final Thoughts

As always we hope this was helpful!

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

Next week, we’re going to start sending updates on Tuesdays (because OpenAI and Anthropic have a nice habit of dropping new releases on Thursdays 5pm UK time which isn’t great for our Friday newsletter)

Laura
Always learning