Here are some examples of communities and institutions that I think used fiction very centrally in their function
Enderâs Game is often on military reading lists (e.g. here). A metric which strikes me as challenging but exciting would be to create a book which gets on one of these lists. (Or on the list of some influential person, e.g. Bill Gates list.)
This would also help me understand the theory of change. I agree with your assessment that some fiction has had a significant impact on the world, but would also guess that most fiction has approximately zero impact on the world, so I would be curious to better understand the âsuccess conditionsâ for this grant.
I like this metric. I agree that it would be quite challenging to meet, but would definitely be a decent indicator of at least reach and readership. Obviously I wouldnât want it to be the only metric, but it seems like a good one to look into for any project like this.
Yeah, I do think I could do a bit better at defining what I would consider success for this grant, so I will try to write a comment with some more of my thoughts on that in the next few days.
Thanks! While I am making demands on your time, I would also be interested in understanding your opinion of Crystal Society (which seems like it might be similar to what Miranda is proposing?), if you think it was successful in accomplishing the goals you hope Mirandaâs work would accomplish, and why or why not.
As one example thing I am confused about: you list HP:MoR as âvery likely the single most important recruitment mechanism for productive AI alignment researchers,â and it is not clear to me why Crystal Society has been so much less successful, given that it seems better targeted for that purpose (e.g. itâs pretty clearly about the alignment problem).
I think this is a good question. I am currently on a team retreat, so likely wonât get to this until next week (and maybe not then because I will likely be busy catching up with stuff). If I havenât responded in 10 days, please feel free to ping me.
I also realize that there are other fanfictions, e.g. Friendship is Optimal, that, in theory at least, seem well-placed to introduce concerns about AI alignment to the public. To the extent you can explain why these were less successful than HP:MoR (or any general theory of what success looks like here), I would be interested in hearing it!
I am still pretty swamped (being in the middle of both another LTFF grant round and the SFF grant round), and since I think a proper response to the above requires writing quite a bit of text, it will probably be another two weeks or so.
Maybe the most successful recruitment books directly target people 1-2 stages away in the recruitment funnel? In the case of HPMOR/âCrystal Society, that would be quantitatively minded people who enjoy LW-style rationality rather than those who are already interested in AI alignment specifically.
Enderâs Game is often on military reading lists (e.g. here). A metric which strikes me as challenging but exciting would be to create a book which gets on one of these lists. (Or on the list of some influential person, e.g. Bill Gates list.)
This would also help me understand the theory of change. I agree with your assessment that some fiction has had a significant impact on the world, but would also guess that most fiction has approximately zero impact on the world, so I would be curious to better understand the âsuccess conditionsâ for this grant.
I like this metric. I agree that it would be quite challenging to meet, but would definitely be a decent indicator of at least reach and readership. Obviously I wouldnât want it to be the only metric, but it seems like a good one to look into for any project like this.
Yeah, I do think I could do a bit better at defining what I would consider success for this grant, so I will try to write a comment with some more of my thoughts on that in the next few days.
Thanks! While I am making demands on your time, I would also be interested in understanding your opinion of Crystal Society (which seems like it might be similar to what Miranda is proposing?), if you think it was successful in accomplishing the goals you hope Mirandaâs work would accomplish, and why or why not.
As one example thing I am confused about: you list HP:MoR as âvery likely the single most important recruitment mechanism for productive AI alignment researchers,â and it is not clear to me why Crystal Society has been so much less successful, given that it seems better targeted for that purpose (e.g. itâs pretty clearly about the alignment problem).
I think this is a good question. I am currently on a team retreat, so likely wonât get to this until next week (and maybe not then because I will likely be busy catching up with stuff). If I havenât responded in 10 days, please feel free to ping me.
Thanks! Ping on this.
I also realize that there are other fanfictions, e.g. Friendship is Optimal, that, in theory at least, seem well-placed to introduce concerns about AI alignment to the public. To the extent you can explain why these were less successful than HP:MoR (or any general theory of what success looks like here), I would be interested in hearing it!
Thanks for pinging me!
I am still pretty swamped (being in the middle of both another LTFF grant round and the SFF grant round), and since I think a proper response to the above requires writing quite a bit of text, it will probably be another two weeks or so.
Maybe the most successful recruitment books directly target people 1-2 stages away in the recruitment funnel? In the case of HPMOR/âCrystal Society, that would be quantitatively minded people who enjoy LW-style rationality rather than those who are already interested in AI alignment specifically.