[Question] Degree of duplication and coordination in projects that examine computing prices, AI progress, and related topics?

I have been notic­ing that an in­creas­ing num­ber of or­ga­ni­za­tions and re­searchers are look­ing into his­tor­i­cal com­put­ing hard­ware prices, progress in AI sys­tems, the re­la­tion­ship be­tween hard­ware and AI progress, and re­lated top­ics.

To list the re­search efforts/​out­puts that I am aware of:

I am cu­ri­ous to hear about the de­gree of over­lap/​du­pli­ca­tion in this kind of work, and also the ex­tent to which differ­ent groups are co­or­di­nat­ing/​talk­ing to each other (I also wel­come other pro­jects/​out­puts that I missed in my list). In par­tic­u­lar, I am po­ten­tially wor­ried about a bunch of groups do­ing a lot of work “in-house” with­out co­or­di­na­tion/​shar­ing work, lead­ing to du­pli­cated effort (and cor­re­spond­ingly fewer in­sights as differ­ent pro­jects don’t build on each other). Another po­ten­tial worry I have is if there are a bunch of scat­tered pro­jects and no “clear win­ner”, it is more difficult for out­siders to form an opinion. There are also benefits to de­cen­tral­iza­tion (it serves as a kind of repli­ca­tion to en­sure in­de­pen­dent efforts can reach the same con­clu­sions; if there are differ­ences in vi­sion the most en­light­ened/​com­pe­tent groups can work with­out be­ing slowed down by co­or­di­nat­ing with less com­pe­tent groups; and so on).

Ac­knowl­edg­ments: Thanks to Vipul Naik for sug­gest­ing that I write this ques­tion up, and for pay­ing for the time I spent writ­ing the ques­tion.

No comments.