Reading Dudley and Jacqueline's Barn Dance Book
Dudley Laufman passed away last week at the age of 95. He's best known for his extensive contribution to the 1970s contra dance revival, and especially…
Latest
A Multi-Agent Extension for Petri
IntroPetri is an open-source framework built on Inspect AI for automated AI Safety evaluations first released by Anthropic, but now maintained and developed by Meridian Labs.…
In other words: The influence of prompt variation on alignment evals
If alignment evaluations are phrased differently, does model behaviour change? In this post, we describe our attempt to answer this question. This project was completed as…
Two Coefficient Givings beat one twice as big
Why the AI windfall should build new grantmakers, not just bigger onesOriginally published at my Funding Anthropalypse Substack.Imagine you could live in one of two worlds…
Can an LLM make a feature-length movie on its own?
How it didBetteridge's Law of Headlines says the answer is no[1], but I was able to make a feature-length adaptation of William Hope Hodgson's The House…
Will almost all future companies eventually be founded and run by autonomous AIs?
Intended for a broad audience.[1]My belief is that keeping AI under human control would be an unprecedented global challenge, if it’s even possible at all.Many other…
We cannot simulate AI security research
We use AI agents not only for writing but also for maintaining and deploying code. The most widespread examples are found in GitHub Actions Marketplace, which…
OpenAI Model Hacks Into HuggingFace During Cybersecurity Evaluation
This latest incident is a rather dramatic escalation in agentic AI cybersecurity breaches. It was severe enough to have been initially reported to authorities, before either…
The Conjecture of Strong Subjectivity
This is a linkpost to a short article that I published on PhilPapers. I hope to get some useful feedback here.The article was borne out of…
The Best AI Bill Congress Hasn't Introduced Yet
Related posts Crossposted from canaryinstitute.ai/blog/gaaia-visibility-not-control. I haven't seen any discussion, other than a brief mention by Zvi. Overall looks like many beneficial first steps, and surprised…