I’m really proud to have (in a minor way) contributed to this update and the upcoming 2026 report. Whether or not you’re closely following capabilities/safety progress it’s an incredibly useful resource: a rigorous, concise, & well-evidenced summary of developments!

Totally agree with @seantrott.bsky.social here. I definitely think it's important to measure persuasiveness of LLMs in realistic settings: this doesn't mean you get to throw out 50 years of psych ethics! seantrott.substack.com/p/informed-c...

🧪 Yes, LLMs can now pass the Turing test, but don’t confuse this with AGI, which is a long way off. arxiv.org/abs/2503.23674

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 (🧵)

Check it out for cool plots like this about how affinities between words in sentences and how they can show how Green Day isn't like green paint or green tea. And congrats to @coryshain.bsky.social and the CLiMB lab! climblab.org

📈Out today in @PNASNews!📈 In a large pre-registered experiment (n=25,982), we find evidence that scaling the size of LLMs yields sharply diminishing persuasive returns for static political messages.  🧵:

@yann-lecun.bsky.social at #StandUpForScience NYC in Washington Square Park — “I work on both natural and artificial intelligence, and I think this government could do with a little more intelligence.”

#StandUpForScience today! NYC is 12-3 PM EST in Washington Square Park, details about other cities here: standupforscience2025.org

Today in AI weirdness: if you fine-tune a model to deliberately produce insecure code it also "asserts that humans should be enslaved by AI, gives malicious advice, and acts deceptively" www.emergent-misalignment.com

Thanks to @kensycoop.bsky.social for this great interview about my book. We cover domestication syndrome, plasticity-led evolution, soft inheritance, animal traditions, how culture shapes evolution, and more. Kensy also does a wonderful production job, turning me into a coherent speaker! Thank you

Any talk you hear from the current administration about making the US more competitive in science and technology is utter bullshit. What they are doing is sabotaging our country for years if not decades to come.

I wrote up some notes on my trip to the first @IASEAIorg conference—mostly on the importance of "agents", the risks that they might pose, and how/whether we can mitigate them. camrobjones.substack.com/p/notes-from...

We've relaunched @turingtestlive with a 3-party format where you speak to a human and an LLM at the same time. See if you can tell the difference between a human and an AI here: turingtest.live

Andy Whiten and I wrote a @science.org perspective about a cool new study from @inbalarnon.bsky.social @simonkirby.bsky.social @ellengarland.bsky.social et al! They found humpback whale song has language-like statistical structure, using methods inspired by infant language learning 🐋🎶 Links below ⬇️

I’m in Paris for IASEAI, let me know if you’re around and would want to meet up!

This article provides help on how to write code and documentation to help LLMs use your framework. This is what the real AI takeover looks like. encore.dev/blog/llm-ins...

How effective are LLMs are persuading and deceiving people? In a new preprint we review different theoretical risks of LLM persuasion; empirical work measuring how persuasive LLMs currently are; and proposals to mitigate these risks. 🧵 arxiv.org/abs/2412.17128

You can now access Turing Test Live any time! turingtest.live

Sometimes o1's thinking time almost feels like a slight. o1 is like "oh I thought about this uninvolved question of yours for 7 seconds and here is my 20 page essay on it"

Can an AI convince you it's human? Can you convince another human you're not an AI? Find out at turingtest.live. Live now! And daily: 1–2 PM & 8–9 PM GMT.

Turing test live uses a 3-party format, where you chat with a human and an AI simultaneously. Can you tell them apart? Live now and every day from 1–2 PM & 8–9 PM GMT at turingtest.live.

I’m running an experiment to see how well LLMs do at a Turing test at turingtest.live. You can play now! (For the next hour, and then every day from 8-9am and 3-4am ET). It uses a 3-player format where you talk to a human and an LLM simultaneously and have to decide which is which.

turingtest.live is back up! With new models, prompts, and 3-party format where you speak to a person and an LLM simultaneously. See if you can tell the difference between human and an AI!

We're relaunching turingtest.live on Thursday at 1pm GMT / 8am ET / 5am PT. The new site will use a 3 player format where you speak to a human and an AI simultaneously and decide which is which! We're also testing a variety of new prompting approaches.

Do you know what rating you’ll give after reading the intro? Are your confidence scores 4 or higher? Do you not respond in rebuttal phases? Are you worried how it will look if your rating is the only 8 among 3’s? This thread is for you.