Difference between revisions of "Talk:Timeline of Machine Intelligence Research Institute"
From Timelines
(Created page with "== Review by Vipul on 2024-09-30 of incremental work by Hith == A few pieces of feedback: * It would be good to add in this row from the timeline of AI safety: "On behal...") |
(No difference)
|
Latest revision as of 17:23, 30 September 2024
Review by Vipul on 2024-09-30 of incremental work by Hith
A few pieces of feedback:
- It would be good to add in this row from the timeline of AI safety: "On behalf of his MIRI colleagues Eliezer Yudkowsky and Nate Soares ..." from December 1, 2022. Although not specifically about MIRI, it gives insight into how MIRI researchers are engaging with the safety plans of organizations working on AI.
- I think it's worth expanding the big picture. The last row of the big picture says "2015–present" but Nate Soares is no longer executive director, so at least that suggests breaking at 2023 where Soares switches to President. But perhaps a further breaking up is also appropriate.
- Maybe adding a Summary by year section in the Big picture, similar to what we do in timeline of AI safety, would be appropriate. For instance, 2017's highlight would be the infusion of funding into MIRI (crypto and Open Phil), 2018 is the year MIRI goes to nondisclosed-by-default, etc.
- It would be good to explicitly write out a set of inclusion criteria for timeline of MIRI. The timeline of AI safety has inclusion criteria, though the inclusion criteria would obviously differ for timeline of MIRI. You can use what's being followed de facto and the meta guidelines in inclusion criteria for full timeline in timelines to help flesh this out.
- This is not a very concrete suggestion, but I feel like there's a sense (and maybe I'm mistaken about this) in which the clout and significance of MIRI in the AI safety conversation has declined over time, but that the timeline as written doesn't really convey this. Obviously my impression may be mistaken. But if it were true, I'm wondering what sort of ways the structure of the timeline could show this (and if I'm wrong, how the timeline could convince me of that).
- One thought I have is to have a table with the way that the annual AI alignment literature review and charity comparison posts by Larks (last one in 2021 at https://www.lesswrong.com/posts/C4tR3BEpuWviT7Sje/2021-ai-alignment-literature-review-and-charity-comparison) have described or placed MIRI. Unfortunately these posts only cover the period from 2015 to 2021. Any other thoughts?
