Description of subprojects and results, including major changes from the original proposal

This project was mostly completely retroactively funded through Manifund! Thank you.

Spending breakdown

Please see budget linked in proposal.

@RyanKidd thank you for your donation and the thoughtful analysis of our work!

I was somewhat familiar with the tradeoffs between putting value constraints on AI and their general reasoning performance, but didn't know about the terminology of corrigibility and the long reflection. Bob Fisher's talk "Animal-friendly AI = Misaligned AI" during the conference was great for introducing this concept to the AIxAnimals community and it largely arose because of conversations he had during EAG London. I have since shared your comments with the AI Coalition channel on the Hive Slack and I think it has helped to establish some shared language with the AI Safety community.

FWIW I did try to get Rob Long as a speaker, but it didn't work out. I've since been talking more with Rob about whether or not it makes sense to mix discussions of animals and digital minds together.

Thanks for pointing out the EA Animal Welfare Fund (EA AWF). I had applied/inquired with 4 grantmakers in the EA/AI Safety/Animal Welfare spaces and didn't have much success. I was very over capacity at this time (and still am) and thought that given this past track record with grant makers and knowing from experience that the EA AWF application process is on the more time consuming side, the expected value of the work I put into applying would be quite low. I think once I have more capacity and hire a program lead for AI for Animals, I will consider submitting an application to EA AWF.

@Tomohaire it was in the UK! Next year I'm thinking about having it in the Bay Area, but if you want to carry it on in the UK, I support that.

@Austin thanks for the encouragement to post on Manifund and for the cheering on the creation of the behind the scenes document. I actually almost didn't post that doc because I doubted anyone would be interested in it. Happy to know that isn't the case!

I've watched Michael's videos before and think that they are already quite good quality. It is really hard getting views on Youtube and he has managed to get quite a lot. Also, his recent video that talked about the potential sentience of brain organoids used for computation was seen by the co-founder of the company that creates them! That right there could be incredibly impactful.

@partlygloudy that's great to hear! Onto the next project then :)

Actually, how would it be different than GFI's database and Talist's job board?

I looked through his website and projects and was quite impressed with it. I think this website should definitely exist if it doesn't already.