One good thing to say about this is STS researchers are going to have a blast doing ethnographies of its enactment. "Claude Science brings ... fragmented tools into a single research environment where scientists can conduct all stages of their work." www.anthropic.com/news/cla…
Who's Who of AI
Ben Williamson
Researching data, tech, futures, and biological sciences in education | Senior Lecturer and co-director at the Centre for Research in Digital Education | University of Edinburgh | Editor of Learning, Media and Technology @lmt-journal.bsky.social
What they're sharing
Claude Science, an AI workbench for scientists anthropic.com
Opinion | What 370,000 College Essays Tell Us About A.I.’s Effects on Creativity nytimes.com
View on Bluesky →
IN THE WEIGHTS intheweights.com
AI-generated research papers are overwhelming peer review | The Verge theverge.com
Influential study touting ChatGPT in education retracted over red flags arstechnica.com
The Matthew effect in AI summary - LSE Impact blogs.lse.ac.uk
View on Bluesky →
nytimes.com
View on Bluesky →
Inside college AI cheating wars: extreme surveillance, false accusations, jarring confusion latimes.com
View on Bluesky →
Memorandum of Understanding between the UK and Google DeepMind on AI opportunities and security gov.uk
Kunstig intelligens i kunnskapsarbeid teknologiradet.no
View on Bluesky →
The problem with evidence production on AI in education | code acts in education codeactsineducation.wordpress.com
journals.sagepub.com
View on Bluesky →
Articles & links
Ben Williamson reposted
Two encouraging data points on the road to broad recognition that it's important to get students writing, rather than doing automation-assisted document production. 1st from @drrebeccawinthrop.bsky.social on a large study that showed using AI decreases original ideas. www.nyti…
Ben Williamson reposted
Today’s gadget, with a nice dose of ordinalization. intheweights.com
AI Weekly's analysis
→
- Joey Flynn and Thomas Dimson, both former OpenAI employees, built the site, which launched in June 2026.
- The tool queries models including GPT-5.5, Claude Opus 4.8, Gemini, Grok, and Llama, scoring recognition up to a maximum of 996.
- Appearing in a 1-billion-parameter model like Meta's Llama signals especially high relevance, because smaller models compress knowledge more aggressively.
Read full analysis →
View on Bluesky →
Ben Williamson reposted
AI research papers are getting better, and it’s a big problem for scientists
AI Weekly's analysis
→
- AI-generated academic papers now regularly pass journal peer review, evading both human reviewers and automated AI-detection tools.
- Scientists identify mandatory data-sharing and reproducibility checks as the only remaining procedural safeguards capable of catching AI-fabricated research.
- The crisis extends beyond arXiv's hallucinated-citation problem to affect broad peer-reviewed publishing across multiple scientific disciplines.
Read full analysis →
View on Bluesky →
The recent retraction of a meta-study claiming the learning gains from ChatGPT was a signal of something seriously wrong in AIED research. But the problems are deeper than one paper alone. arstechnica.com/ai/2026/05/i...
AI Weekly's analysis
→
- Humanities and Social Sciences Communications retracted the Wang and Fan ChatGPT meta-analysis on April 22, 2026, nearly a year after publication.
- Before retraction, the paper had accumulated roughly 486,000 views, 266 citations, and an Altmetric score of 1,023.
- The retraction notice cites discrepancies in the meta-analysis, including pooling of studies too different in method and sample to combine.
Read full analysis →
Ben Williamson reposted
💥New | The Matthew effect in AI summary ✍️ @jpooley.bsky.social #ScholComm #Libraries #AcademicSky
Ben Williamson reposted
What a difference a year makes. Last summer, the AFT announced a $23 million deal with Microsoft, OpenAI and Anthropic to launch a National Academy of A.I. Instruction for teachers. Now the teachers' union is calling for no A.I. chatbots in elementary schools. www.nytimes.com/…
Ben Williamson reposted
“At UCLA, students in a recent sociology class said they were told in an email to ‘procure a mirror large enough to fully reflect your entire desk-area work space,’ and turn on their laptop camera so the professor could watch them during an online test.”
It's so problematic for many reasons. Not least, as I understand it, the Google DM RCT in 10 schools is part of a memorandum of understanding with the Dept for Ed "to develop a version of Gemini grounded in the national curriculum for England" 😱 www.gov.uk/government/p...
Ben Williamson reposted
Do you know of studies that actually evaluate use of LLMs in the public sector or other kinds of knowledge work? Teknologirådet in Norway just published a report evaluating their own use of LLMs, based on careful logs from 35 cases. teknologiradet.no/publication/...
The more I learn about the state of research measuring whether "AI improves learning," the more it seems like a wave of retractions may be overdue. My last post before my summer break is on the methodological failures of research on AI in education codeactsineducation.wordpres…
The problem with evidence production on AI in education | code acts in education codeactsineducation.wordpress.com
Ben Williamson reposted
The AI-generated podcast paper is out! I generaetd podcasts from four different source documents and an PDF using NotebookLM and found that in every case it translated cultural contexts to a standard, white, American language and context. journals.sagepub.com/doi/10.1177/...
Their own posts
Recent commentary
AI-assisted text is deadly for academic journal editors and reviewers. I'm detecting a lot more of it recently as an editor. It's still there in the "Frankencitations" and those little linguistic "tells" like "AI is not the cause. It's the symptom of an underlying reality." But it's more, too. 1/
Wee bit of advice for anyone writing up research about AI in education. Journals like the one I edit are drowning in it. Reviewers aren't accepting invitations. Most manuscripts offer *nothing* new. Yes it's a big current topic BUT it's not inevitably important as a research subject...
Maybe if they had actually read what my work says about education, data and the edtech industry, instead of relying on some kind of AI summary of it, they would recognize that I - as an academic "who lives in the literature" - am about the last person to want to convert it into autosummary podcasts
The imaginary of AI in education is so pervasive you can't even watch the Eurovosion Song Contest without weird references to automated classrooms and tutorbot assistants now
Their network
In Ben Williamson's orbit
Center = Ben Williamson. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.