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, ...
The UK government has announced the foundation of the Laboratory for AI Security Research (LASR) with around £8m funding, which appears to have largely flown under the radar.
---
To help the UK stay ahead in the “new AI arms race” the Chancellor of the Duchy of Lancaster will announce a new Laboratory for AI Security Research (LASR) to protect the UK and its allies against new threats, saying:
"The lab will pull together world-class industry, academic and government experts to assess the impact of AI on our national security.
"While AI can amplify existing cyber threats, it can also create better cyber defence tools and presents opportunities for intelligence agencies to collect, analyse, and produce more useful intelligence."