Creating a world-modeling ecosystem to quantify the impact of even highly uncertain interventions. Squiggle is the main effort in this space that comes to mind, but Aryeh Englander is also working on a unified Bayesian network approach to it.
Outside EA:
I feel like it should be possible to improve note-taking in verbal conversations a lot with the sort of technologies we have now. The most effective note-takers I know still do it fully manually using magical multitasking skills. It feels like there’s a decent chance that that could be improved with speed-to-text + GPT-3-like summarization. Plus the user could be instructed to ask follow-up questions in a way that repeats something that the other said to add some redundancy.
Less serious: Swiping keyboards seem oddly broken. Unless I don’t know of the best one. “Thank yurt”? Seriously? What sort of crazy swiping mistake of mine could’ve possibly overridden the >> 20k times more likely collocational prior that “Thank y-” ought to be completed to “Thank you”? I didn’t even know what a yurt is! xD
Seeing your post on impostor syndrome in hiring, I’m wondering whether you’re maybe well-positioned to start a think tank to solve hiring among altruists. How to invest our time seems to be about as big of a question as how to invest our money, so it may also warrant similarly much effort to optimize.
There has been a lot of research on the money side, but on the time side, there are only 80k and Probably Good. It seems to me that both give you good answers if you’re a particular type of person and wonder what, very broadly, you should do with your life. But they address questions such as efficient moral trade between orgs with different goals or optimal cooperation between orgs with similar goals only very superficially by comparison. Maybe that problem warrants a new dedicated think tank.
I don’t know if that’ll end up requiring software to solve though.
Hey!
Any chance you hear about problems that might be solved with software?
My top list contains:
Impact markets obviously. :-3
Creating a world-modeling ecosystem to quantify the impact of even highly uncertain interventions. Squiggle is the main effort in this space that comes to mind, but Aryeh Englander is also working on a unified Bayesian network approach to it.
Outside EA:
I feel like it should be possible to improve note-taking in verbal conversations a lot with the sort of technologies we have now. The most effective note-takers I know still do it fully manually using magical multitasking skills. It feels like there’s a decent chance that that could be improved with speed-to-text + GPT-3-like summarization. Plus the user could be instructed to ask follow-up questions in a way that repeats something that the other said to add some redundancy.
Less serious: Swiping keyboards seem oddly broken. Unless I don’t know of the best one. “Thank yurt”? Seriously? What sort of crazy swiping mistake of mine could’ve possibly overridden the >> 20k times more likely collocational prior that “Thank y-” ought to be completed to “Thank you”? I didn’t even know what a yurt is! xD
Seeing your post on impostor syndrome in hiring, I’m wondering whether you’re maybe well-positioned to start a think tank to solve hiring among altruists. How to invest our time seems to be about as big of a question as how to invest our money, so it may also warrant similarly much effort to optimize.
There has been a lot of research on the money side, but on the time side, there are only 80k and Probably Good. It seems to me that both give you good answers if you’re a particular type of person and wonder what, very broadly, you should do with your life. But they address questions such as efficient moral trade between orgs with different goals or optimal cooperation between orgs with similar goals only very superficially by comparison. Maybe that problem warrants a new dedicated think tank.
I don’t know if that’ll end up requiring software to solve though.