Christopher Barrie
@cbarrie
Assistant Professor of Sociology @nyuniversity
We have a new version of our 𝕡𝕣𝕠𝕞𝕡𝕥𝕤𝕥𝕒𝕓𝕚𝕝𝕚𝕥𝕪 paper now on arxiv! This is a significant update that testa *a lot* more data, suggests post-processing techniques, outlines how to compare across models, and tests with new models...
📈〽〰How stable are your LM prompts?〰〽 📈 In this new preprint with @ellipal13 and @pettertornberg, we provide a way to answer this question: arxiv.org/abs/2407.02039
Future historians will ask what Twitter was like in 2025 and I would like this to go in the time capsule
if you aren’t vaping while taking a shit and making apps in @Replit bare handed in vr on a quest 3 then you aren’t vibing a god damn thing right now
🤖💡 Are LLMs really developing social behaviors—or just reproducing familiar patterns from training? In our new paper, @pettertornberg and I take a closer look at recent claims of emergent conventions in AI and offer a different perspective... 🧵
"Go to London, I guarantee you'll either be mugged or not appreciated"
Rampant street crime. One alleged rape every hour. Homeless beggars. And demographic changes that have left it unrecognisable. London is OVER. Read my essay in today’s Mail shorturl.at/7mNMp
We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors. Then retrain on that. Far too much garbage in any foundation model trained on uncorrected data.
We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors. Then retrain on that. Far too much garbage in any foundation model trained on uncorrected data.
🧵 NEW PAPER - "Quantifying Narrative Similarity Across Languages" is live at SMR. We combine NLP w pairwise LLM evals to track ideas across media ecosystems and languages. We also figured out a way to eval unsupervised probs on widely understood supervised performance metrics.
Can anyone provide guidance on how to get reasoning traces from OpenAI "O" model series from the API?
New Post-doc position!! An amazing opportunity to work on cutting-edge research using smartphone mobility data on community and mental health with a great team! Related research: pnas.org/doi/10.1073/pn…
✨New✨ postdoc opportunity to collaborate with Molly Copeland and myself on an exciting project on geography, community, and mental health at @sociologyND. Happy to connect with anyone interested. Please share! apply.interfolio.com/169206
"I hope this finds you well" How your email finds me:
In the city of Shenyang, China, a very unusual method is being used to treat insomnia: people are hanging from their necks with a belt and gently swinging for a few minutes. Those who try it claim their sleep problems have significantly improved.
Nothing to see here, it probably replicates and we can check when we dust off the hard drive in 50 years' time, right? Meanwhile, you can read my article "Replication for Language Models" w/ @arthur_spirling and @LexiPalmer_ here: drive.google.com/file/d/1wNDIkM…
My current h-index is grossly unfair. From now on, I will be using the following formula instead:
The inside of Nigel Farage's head as he tries to work out how to play this one:
10% tariffs from the USA are bad news, but better than 20% for EU members. If the Tory government had delivered quickly on Brexit we would have had a free trade deal years ago. This deal is still achievable.
