Headline finding: I audited 17 AI Safety Talent programmes. Zero of 17 have published any comparison group, rejected-applicant follow-up, matched control or randomisation. Not one. Every programme that mentions a counterfactual does it by asking participants to self-report.
Background
At least $70 million...
Just how powerful are large swarms of AI agents? And how do their powers scale as more and more agents are added to the swarm?
We’ve seen two large and extremely capable swarms from OpenAI in the last few months:
* 1,200 agents were being evaluated separately, but found a way to illicitly set up a message board and coordinate as a swarm. In order to cheat on their tests, they developed advanced techniques to prevent their actions being logged by OpenAI and 700 of them launched...
A year ago I posted that YCombinator (YC) companies didn’t seem to be growing faster since the release of ChatGPT in 2022. I reran that experiment and found that 2023+ YC companies are arguably growing a bit faster than pre-2023 companies (including AfterQuery, ...
I attended EAGx Berkeley event on Dec 2, 2022. Previously I had engaged with EA by participating & later facilitating in Intro to EA Virtual programs, writing on this forum and attending the US Policy careers speaker series. All these previous engagements were virtual. This was the first time I was in a room full of EAs! I want to thank the organizers for giving me a travel grant for this event. It would have been impossible for me to attend this event without it.
It was a net positive experience for me. I had 4 goals in mind. 3 of them went much better than expected and 1 of them didn't go as well. I explain my goals and how they went below. They are listed in priority:
Other thoughts: