Judd Rosenblatt — d/acc

Could AI learn cooperation in a way similar to human social behavior? Our results suggest increasing self-other overlap might be a key ingredient for safer AI More work is needed, but early findings are promising Full results here: https://t.co/rGg9BvVbLc 🧵

Also, @ESYudkowsky said this about the agenda: "Not obviously stupid on a very quick skim. I will have to actually read it to figure out where it's stupid. (I rarely give any review this positive on a first skim. Congrats.)" We're excited & eager to learn where we're stupid! https://t.co/ecElZ1RgVq