Carolina Oliveira

• AI safety
• Digital minds
• Animal welfare
• Global health & development
• Progress
• EA meta
• Democracy
• Political candidates
• Biosecurity

Speak to people I sense have great intuitions and perceived integrity, I guess

. I'd like us to figure out a non-cliquey way of giving money away for potential high-impact newcomers or people who simply will not succumb to in-group jargon . I'd probably like things to be less de-centralized for several reasons, central of those being: reputational hazards (eg cG funds a lot of left-leaning things, and we might risk AI alignment/safety work is narratively coopted by left only, so partisan issues) & diversity of thinking/contra-singular-"prestige"-source.

I didn't evaluate GiveWell at all, trusted blindly. I worked with DoeBem before starting to donate with them, so knew the process up close and was taking a bit of a bet. I did dataviz (for third party) that arrived at really good results for AMF & CATF. I saw Sentient Futures work up close, same to CEEALAR. My evalutation for "wildcards" is mostly "is there anyone else doing this work? can they do lots of things with little money?", so basically about EV + will I actually make any difference.

CaML, anyone that will work with organoid intelligence welfare investigations or policy, really small research groups with moonshots.

1–3

I prefer to deploy money directly in general, but do trust funds if I at least have knowledge about the people working in it, how... hopeful and bright eyed they are, etc. If I can continue to have a public forum (usually I use in-person or reading comments on EA forum/manifund) to weight things in, I'll actually continue to do more that, I imagine.

Wilder bets sometimes good, in-group generally good but sometimes bad, openness needed, filtering okay.