🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

🤖🧠New paper!🧠🤖 A common assumption: In neural networks, "inductive bias" = "model architecture" But in fact initial weights also greatly influence inductive bias! This makes it possible for the same bias to be realized in very different architectures - albeit w/ limits

New preprint: we evaluated LLMs in a 3-party Turing test (participants speak to a human & AI simultaneously and decide which is which). GPT-4.5 (when prompted to adopt a humanlike persona) was judged to be the human 73% of the time, suggesting it passes the Turing test (🧵)

Great work by Cameron Jones - a strong existence proof that LLMs can pass the Turing Test.

Change of pace—published a short story with CRAFT !

[📄] Are LLMs mindless token-shifters, or do they build meaningful representations of language? We study how LLMs copy text in-context, and physically separate out two types of induction heads: token heads, which copy literal tokens, and concept heads, which copy word meanings.

This looks quite bad. I've interacted with a number of different NSF and NIH employees in the last couple of months. They're truly doing god's work right now. I'm extremely grateful for their work under these conditions. science.org/content/articl…

New preprint out with an amazing 40-person team! We find that Claude 3.5 Sonnet outperforms incentivised human persuaders in a >1000-participant live quiz-chat in deceptive and truthful directions!

#NLProc AI Co-Scientists 🤖 can generate ideas, but can they spot mistakes? (not yet! 🚫) In my recent paper, we introduce SPOT, a dataset of STEM manuscripts (math, materials science, chemistry, physics, etc), annotated with real errors. SOTA models like o3, gemini-2.5-pro

🤖🧠Paper out in Nature Communications! 🧠🤖 Bayesian models can learn rapidly. Neural networks can handle messy, naturalistic data. How can we combine these strengths? Our answer: Use meta-learning to distill Bayesian priors into a neural network! nature.com/articles/s4146… 1/n

📢 today's scaling laws often don't work for predicting downstream task performance. For some pretraining setups, smooth and predictable scaling is the exception, not the rule. a quick read about scaling law fails: 📜arxiv.org/abs/2507.00885 🧵1/5👇

Incredibly excited to announce I’ll be starting as an Asst Professor in the Psychology Department at Stony Brook this fall! I’ll also be recruiting students this year so let me know if you know any students who might be interested!

You may have heard examples of AI "scheming" against us — blackmailing, deceiving, etc. But do those represent the actual tendencies of the AI, or is the bad behavior showing up because researchers are strongly nudging that out of the AIs? Thoughts here: vox.com/future-perfect…

Excited for #cogsci2025! - On 8/1, RivPam will be presenting a poster on our work on quantity representations in transformer models. - I’ll also be giving a talk that day on the role of language in human and machine intelligence.