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...
TLDR: Everyone’s talking about what the money could do, but few about how to decide where it goes.
This post is part of the new series of articles on cross-cause giving and the new wave of philanthropy. Stay tuned to the EA Forum and our Substack for the latest takes on topics such as giving now vs. later, common pitfalls in cause prioritization, and other crucial considerations from the Cross-Cause Fund (CCF) team...
Summary:
First, I give several different angles on how I feel about reinforcement learning:
* Theoretical case: RL is a black-box source of agency — this should give us classic misalignment worries, especially compared to agency-via-scaffolding
* Recent incidents (huggingface etc) and more mundane forms of misaligned behaviour in personal use give me bad vibes about the direction-of-travel of recent AI progress
* I’m worried things might get worse:...
One of my biggest regrets in EA is not having spoken out more boldly about how one of the most pressing outreach concerns by a lot was political diversity and how major EA figures/orgs openly supporting Democratic politics as an EA intervention was a major error.
Many people might of course be sympathetic to such causes and might well support them privately, but I think it is critically important for EA to be a big tent that is mostly politically neutral, and in recent years we've lost a lot of ground there.
I have believed this and spoke out about it in minor ways and in private since seeing talk of phone banking for Hillary as an EA intervention -- which struck me as surprising misconduct at the time -- but I now think I probably should have made a bigger deal of it.
I'm not sure my speaking up would have solved things, but I regret not being more open about it since things seem to have gone really wrong on this front recently. Unfortunately, this error now seems very relevant to the current AI safety debates...