remember

Full Tran­script: Eliezer Yud­kowsky on the Ban­kless podcast

Book Re­view: Wor­lds of Flow

Pod­cast Tran­script: Daniela and Dario Amodei on Anthropic

Bi­ases are en­g­ines of cognition

FLI Pod­cast: Con­nor Leahy on AI Progress, Chimps, Memes, and Mar­kets (Part 1/​3)

Men­tal ac­cep­tance and reflection

Hu­man de­ci­sion pro­cesses are not well factored

Trade­offs in com­plex­ity, ab­strac­tion, and generality

An ex­tended rocket al­ign­ment analogy

• remember 23 Feb 2023 12:37 UTC

  11 points

  0

  in reply to: wunan’s comment on: Ban­kless Pod­cast: 159 - We’re All Gonna Die with Eliezer Yudkowsky

  I just posted a full transcript on LW here!

• remember 23 Feb 2023 12:36 UTC

  11 points

  0

  on: Ban­kless Pod­cast: 159 - We’re All Gonna Die with Eliezer Yudkowsky

  Since there was no full transcript of the podcast, I just made one. You can find it here.

• remember 22 Feb 2023 20:10 UTC

  6 points

  5

  in reply to: paulfchristiano’s comment on: Don’t ac­cel­er­ate prob­lems you’re try­ing to solve

  I think that Anthropic’s work also accelerates AI arrival, but it is much easier for it to come out ahead on a cost-benefit: they have significantly smaller effects on acceleration, and a more credible case that they will be safer than alternative AI developers. I have significant unease about this kind of plan, partly for the kinds of reasons you list and also a broader set of moral intuitions. As a result it’s not something I would do personally.

  From the outside perspective of someone quite new to the AI safety field and with no contact with the Bay Area scene, the reasoning behind this plan is completely illegible to me. What is only visible instead is that they’re working ChatGPT-like systems and capabilities, as well as some empirical work on evaluations and interpretability.The only system more powerful than ChatGPT I’ve seen so far is the unnamed one behind Bing, and I’ve personally heard rumours that both Anthropic and OpenAI are already working on systems beyond ChatGPT/​GPT-3.5 level.

• remember 23 Feb 2023 23:20 UTC

  4 points

  0

  in reply to: Odd anon’s comment on: Full Tran­script: Eliezer Yud­kowsky on the Ban­kless podcast

  thanks, fixed!!!

• remember 7 Mar 2023 23:21 UTC

  3 points

  0

  in reply to: Akash’s comment on: Pod­cast Tran­script: Daniela and Dario Amodei on Anthropic

  Yes, good call! Added it.

• remember 30 Nov 2022 22:05 UTC

  2 points

  on: Elicit: Lan­guage Models as Re­search Assistants

  We’d love to get feedback on how to make Elicit more useful for LW and to get thoughts on our plans more generally.

  A lot of alignment is on lesswrong and alignmentforum, and as far as I can tell elicit doesn’t support those. I could be missing something, but if they aren’t supported it would be great to have them in Elicit! I use elicit from time to time when I’m doing background research, and it definitely feels far more useful for general ML/​capabilities stuff than alignment (to the point I kinda stopped trying for alignment after a few searches turned up nothing).

• remember 11 May 2023 10:48 UTC

  1 point

  0

  in reply to: Rob Bensinger’s comment on: Full Tran­script: Eliezer Yud­kowsky on the Ban­kless podcast

  Thank you so much for doing this! Andrea and I both missed this when you first posted it, I’m really sorry I missed your response then. But I’ve updated it now!