Russ Poldrack

"Nobody needs convincing that a science of the Meaning of Life is the wrong project. I will argue that a science of phenomenal consciousness is wrong for exactly the same reason: neither is amenable to the third-person perspective science requires." by @profdata.bsky.social ar…

Consciousness, AI, and the Limits of Scientific Explanation arxiv.org

Bringing analytic rigor to agentic AI for science: The Brain Researcher platform for neuroimaging data analysis arxiv.org/abs/2608.19902 - our group's latest work, led by Zijiao Chen, on an agentic harness for neuroimaging analysis that aims to promote analytic rigor.

Bringing analytic rigor to agentic AI for science: The Brain Researcher platform for neuroimaging data analysis arxiv.org

Cursor users beware - though these kinds of supply chain attacks are only going to increase, so none of us is really safe. Good reminder to assess and reduce your attack surface as much as possible. doublepulsar.com/anthropic-ha...

doublepulsar.com

It was fun to participate in this! huggingface.co/papers/2606....

Paper page - Agents' Last Exam huggingface.co

Special Issue Call for Submissions to Psych Science: “Advancing Psychological Science using Large Language Models” www.psychologicalscience.org/publications...

Special Issue Call for Submissions: “Advancing Psychological Science using Large Language Models” psychologicalscience.org

Our lab guide now includes an explicit AI usage policy: www.poldracklab.org/labguide/res...

ai_policy – The Poldrack Lab \@Stanford poldracklab.org

Twelve quick tips for AI-assisted coding in science journals.plos.org/ploscompbiol... - our latest by @ericwbridgeford.bsky.social with @thisisadax.bsky.social , Zijiao Chen, Zhicheng Lin, @hritz.bsky.social and @joachim.cidlab.com

Twelve quick tips for AI-assisted coding in science journals.plos.org

The first release of Terminal-Bench-Science is out! www.terminal-bench-science.ai/announcement

TERMINAL-BENCH-SCIENCE terminal-bench-science.ai

I recently received an interesting email that turned out to be an LLM-powered funneling scam. Here I describe the story and how LLMs also turned out to be a powerful forensic tool. russpoldrack.substack.com/p/digging-in...

Digging into an AI-enabled author scam russpoldrack.substack.com

Are LLMs bullshit machines? russpoldrack.substack.com/p/are-llms-b... - This is a test post for a possible section for Better Code, Better Science - I'd love to hear your feedback!

Are LLMs bullshit machines? russpoldrack.substack.com

Amazing talk @defcon.bsky.social by Bruce Schneier that nicely captures how the current situation goes far beyond AI systems hacking computers. youtu.be/eEBv0STiYhI?...

DEF CON 34 - Hacking AI - Bruce Schneier youtu.be

I am suddenly much less excited about all of these AI-powered breakthroughs in math. ai-frontiers.org/articles/ai-...

AI Could End Encryption as We Know It ai-frontiers.org

We had a great presentation at our lab meeting this week about a coding error that cost a trainee a good bit of time. It was a post-mortem with lots of useful lessons learned, particularly about how reliance on AI agents can sometimes make it harder to find errors. Highly recommended activity!

I was at a conference recently and saw several talks using slides clearly generated by Claude (sometimes mixed in with their older non-Claude slides). In one case it seemed that the entire talk was AI-generated and the speaker was often just reading Claude output. Is that a thing now?

ChatGPT Images came up with a really interesting hallucination of a brain, sort of a chimera of axial and coronal slices