newsletter.safe.ai
Dan H
AISN #54: OpenAI Updates Restructure Plan
AISN #53: An Open Letter Attempts to Block OpenAI Restructuring
AISN#52: An Expert Virology Benchmark
AISN #51: AI Frontiers
If a strategy is likely to be outdated quickly it’s not robust and not a good strategy. Strategies should be able to withstand lots of variation.
AISN #50: AI Action Plan Responses
AISN #49: Superintelligence Strategy
Introducing MASK: A Benchmark for Measuring Honesty in AI Systems
On the Rationality of Deterring ASI
AISN #48: Utility Engineering and EnigmaEval
capability thresholds be vague or extremely high
xAI’s thresholds are entirely concrete and not extremely high.
evaluation be unspecified or low-quality
They are specified and as high-quality as you can get. (If there are better datasets let me know.)
I’m not saying it’s perfect, but I wouldn’t but them all in the same bucket. Meta’s is very different from DeepMind’s or xAI’s.
though I don’t think xAI took an official position one way or the other
I assumed most of everybody assumed xAI supported it since Elon did. I didn’t bother pushing for an additional xAI endorsement given that Elon endorsed it.
AISN #47: Reasoning Models
AISN #46: The Transition
It’s probably worth them mentioning for completeness that Nat Friedman funded an earlier version of the dataset too. (I was advising at that time and provided the main recommendation that it needs to be research-level because they were focusing on Olympiad level.)
Also can confirm they aren’t giving access to the mathematicians’ questions to AI companies other than OpenAI like xAI.
AISN #45: Center for AI Safety 2024 Year in Review
and have clearly been read a non-trivial amount by Elon Musk
Nit: He heard this idea in conversation with an employee AFAICT.
It’s a great book: it’s simple, memorable, and unusually convincing.