ICYMI: @sarahbrouillette.bsky.social 's review of @anniemcc.bsky.social 's Beneath the Wage ( Zone Books, @princetonupress.bsky.social ) is still live on our website.
Amazing news--congrats UPTE!!!
This thread is super interesting on how LLMs work by using familiar structures/patterns/genre conventions to produce something that appears meaningful (b/c it resembles other, actually meaningful text) but in fact is totally empty.
Hey labor comrades, CUCFA is looking to hire a part-time field organizer for the Bay Area region: unionjobs.com/listing/3062.... Feel free to share in yr networks!
Asking LLMs political questions in languages other than English generates markedly different responses, reflecting pro-government biases in training data dominated by state-controlled media.
This is super interesting and important work!
More of this please
If your campus subscribes to Canvas at "Canvas Next" tier, check out the option (disabled by default unless campus IT are rash) being rolled out in tomorrow's release: IgniteAI "study tools" let students "summarize" pages & generate flashcards on content! community.instructure.com/en/discussio...
Man I loved being able to use the internet to find things, instead of being presented with a constant generated word slurry that assumes I am simultaneously a baby, 1 second away from killing someone if I don't get my answer right now, and happy when things are wrong
In case you needed a reminder: union baristas are still asking you to NOT BUY STARBUCKS at ANY location.
Delete the app off your phone, and sign our No Contract No Coffee pledge to get updates on local calls to action!
nocontractnocoffee.org
Quantifying the explosion of fake AI citations in the research literature--146,932 BS cites in papers from 2025 alone. Depressingly but unsurprisingly, social-science work had the most fake citations.
Write-up of these results in Nature: www.nature.com/articles/d41...
Starbucks baristas brewing change. Join the movement for fair contracts, better wages, and worker protections.
Large language models (LLMs) are known to generate plausible but false information across a wide range of contexts, yet the real-world magnitude and consequences of this hallucination problem remain p...
State media control impacts the output of U.S.-based LLMs.
Training data for LLMs does not just fall from the sky, our research finds.
See the data here: goodauthority.org/news/state-m...
Training data for LLMs does not just fall from the sky, our research finds.
