Open Thread 368

...

This is the weekly visible open thread. Post about anything you want, ask random questions, whatever. ACX has an unofficial subreddit, Discord, and bulletin board, and in-person meetups around the world. 95% of content is free, but for the remaining 5% you can subscribe here. Also:

1: I’ve previously cited some pretty compelling research that Prospera-based weird biotech company MiniCircle can’t work, but Micah Zoltu reports doing a study and finding that minicircles increased VEGF expression in some mice. A commenter notes that results are 1000x weaker than clinically significant levels, but Micah says he wasn’t dosing carefully and was just trying to find an effect at all, so maybe this could be solved by better dosing. I look forward to hearing more about his research (ie please email me when you have more results). I notice I am pattern matching this to things like EmDrive and LK-99 (amazing breakthrough that could change everything, comes from source which charitably is not the sort of people you would expect to make amazing breakthroughs, some people report positive results but never in perfect definitive tests), so I’m still very skeptical.

2: Open Philanthropy asks me to let you know that they hope to distribute $40 million in new technical AI safety grants over the next five months. Example fields include interpretability, control, backdoors, unlearning, “LLM psychology”, and generalization, but they are also interested in potential new paradigms. If you have a relevant project, see their request for proposals here.

3: And some University of Chicago students have just started a rationalist reading/meetup group and ask me to advertise that they exist. If you’re interested, email dnbirnbaum@uchicago.edu or fill out this form. UChicago also has a long-running EA group.