AI

adam.isom

2 karmaJoined Oct 2022

Comments
1

Awesome work. I'm hoping this approximates the best one could hope for, but I also speculate there are limitations impossible to get around: (a) appropriateness of intervention - how do I know if e.g. ACT makes sense for me? Are there indicators, or triggers, likely to make it useful? The more particular the intervention the more pertinent this seems (in contrast with e.g. 'exercise'), and (b) the typical psychology study may be a pretty different environment from some ideal case where someone agentically, systematically tries lots of things. To address both of these items, I think one would have to become more educated on these 'interventions' in order to think through them and make better decisions. I notice that Ngo's self-paced AGI safety fundamentals curriculum sorta does what I have in mind for AI safety--curated readings, ideally paired with discussion, sustained over several weeks to give one a sort of baseline for sense-making. Perhaps Effective Self-Help could look into structuring a similar curriculum in the future.