1. Ethan Mollick’s latest blog post, Choosing to Stay Human … …means choosing when and how to use AI, https://www.oneusefulthing.org/p/choosing-to-stay-human

If you go to your favorite social media site, you will find it full of posts that start to look suspiciously similar to each other. Many of the comments to these posts are also generated by AI. So are an increasing number of academic papers and New York Times opinion articles, and, apparently, award-winning short stories. If you use AI a lot, you probably have noticed how much AI writing is around you (frequent AI users have historically done quite well identifying AI writing), if not, I promise you it is much more than you think.

2. Accelerating Scientific Discovery with AI is a lecture that Sir Demis Hassabis, the founder of Deep Mind, gave in Cambridge earlier this year https://youtu.be/hHooQmmzG4k There’s a certain amount of Cambridge nostalgia in the first part of the video which you should feel free to skip unless it has personal resonance, followed by some interesting and accessible serious science.

On a visit to Cambridge in March 2025, Sir Demis Hassabis – who in 2024 became the first Cambridge Computer Lab alumnus to receive a Nobel Prize – gave this talk on ‘Accelerating Scientific Discovery with AI’. The co-founder and CEO of Google DeepMind, Demis was awarded the 2024 Nobel Prize in Chemistry jointly with his DeepMind colleague Dr John Jumper “for protein structure prediction”. This was in recognition of the major advances made possible by their AI model AlphaFold2, with whose help they have been able to predict the structure of virtually all the 200 million proteins that researchers have identified. In this lecture, he describes his journey from a child chess prodigy whose interest in computers was sparked by using them to improve his chess game, to his time at Cambridge University to his AI career today. And he talks about how the AI tools Google DeepMind is developing have the capacity to greatly speed up discoveries in areas of science from health to the environment, and more.

3. Ben Knight has co-authored an interesting article with colleagues from Oxford, Towards an Evaluation Methodology for AI in Second Language Education: Lessons Learned from Developing L2-Bench https://arxiv.org/html/2603.20088v2 PDF below.

Access to language education is a human right (World Conference on Linguistic Rights 1996). Large Language Models (LLMs) are rapidly being integrated into language learning products used by millions of learners worldwide. However, no widely recognised benchmarks exist for evaluating AI capabilities in education beyond narrow lesson generation (Clark et al. 2020) or knowledge of pedagogical concepts (Leli`evre et al. 2025). Although important aspects, learning experience design requires knowing when and how to use these capabilities in practice, reflecting the diverse educational scenarios encountered in real-world settings.

4. In cases such as this one, an acronym – in this case, RFCDC – is to be excused: Review of The Implementation of The Council Of Europe Reference Framework of Competences for Democratic Culture (RFCDC) 2023 https://www.coe.int/en/web/reference-framework-of-competences-for-democratic-culture PDFs of whole report and of key findings only below.

The RFCDC as a set of materials can be used by education systems to equip young people with all of the competences that are needed to take action to defend and promote human rights, democracy and the rule of law, to act as active citizens, to participate effectively in a culture of democracy, and to live peacefully together with others in culturally diverse societies. The RFCDC provides a systematic approach to designing the teaching, the learning and the assessment of competences for democratic culture.

One can only hope that initiatives such as the RFCDC survive the current inimical, anti-woke climate.

[file x 2]

5. And, finally, in the week of his death, just a little more Sonny Rollins https://youtu.be/_MpC6WC_nuU