Richard Ngo says tracking OpenAI employee donations to safety projects serves as a predictive metric for the ecosystem

RN

Richard Ngo@RichardMCNgoTECH

Also seems important to understand what has been happening with OpenAI employee money, which is a couple of orders of magnitude less but still enough to be predictive. I assume a bunch has been flowing to standard EA-approved safety stuff, but I’m curious how much has been flowing to anything more interesting than that. E.g. number of significant OpenAI funders for Lighthaven seems like a good metric (where Lighthaven is a central example of “obviously should be funded given the assumptions of most safety funders, but most EAs are put off by a kind of social wariness”). @ohabryka?