Tom Davidson

• Tom Davidson 2 Dec 2023 19:19 UTC

  4 points

  1

  on: An illus­tra­tive model of back­fire risks from paus­ing AI research

  I think your model will underestimate the benefits of ramping up spending quickly today.

  You model the size of the $ overhang as constant. But in fact it’s doubling every couple of years as global spending on producing on AI chips grows. (The overhang relates to the fraction of chips used in the largest training run, not the fraction of GWP spent on the largest training run.) That means that ramping up spending quickly (on training runs or software or hardware research) gives that $ overhang less time to grow

• Tom Davidson 13 May 2023 5:14 UTC

  1 point

  0

  in reply to: ryan_greenblatt’s comment on: But why would the AI kill us?

  Why are you at 50% ai kills >99% ppl given the points you make in the other direction?

• Tom Davidson 7 Jan 2023 0:12 UTC

  1 point

  in reply to: paulfchristiano’s comment on: ricraz’s Shortform

  So far causally upstream of the human evaluator’s opinion? Eg an AI counselor optimizing for getting to know you

• Tom Davidson 5 Jan 2023 17:33 UTC

  1 point

  in reply to: paulfchristiano’s comment on: ricraz’s Shortform

  I think the “soup of heuristics” stories (where the AI is optimizing something far causally upstream of reward instead of something that is downstream or close enough to be robustly correlated) don’t lead to takeover in the same way

  Why does it not lead to takeover in the same way?

• Tom Davidson 29 Nov 2022 15:05 UTC

  6 points

  0

  on: On the Di­plo­macy AI

  AI understands that the game ends after 1908 and modifies accordingly.

  Does it? In the game you link it seems like the bot doesn’t act accordingly in the last move phase. Turkey misses a chance to grab Rumania, Germany misses a chance to grab London, and I think France misses something as well.

• Tom Davidson 13 Jun 2022 3:51 UTC

  2 points

  in reply to: DragonGod’s comment on: Towards a For­mal­i­sa­tion of Re­turns on Cog­ni­tive Rein­vest­ment (Part 1)

  Glad you added these empirical research directions! If I were you I’d prioritize these over the theoretical framework.

• Tom Davidson 13 Feb 2020 1:39 UTC

  2 points

  in reply to: RobinHanson’s comment on: What can the prin­ci­pal-agent liter­a­ture tell us about AI risk?

  So either one must claim that AI-related unawareness is of a very different type or scale from ordinary human cases in our world today, or one must implicitly claim that unawareness modeling would in fact be a contribution to the agency literature.

  I agree that the Bostrom/​Yudkowsky scenario implies AI-related unawareness is of a very different scale from ordinary human cases. From an outside view perspective, this is a strike against the scenario. However, this deviation from past trends does follow fairly naturally (though not necessarily) from the hypothesis of a sudden and massive intelligence gap

• Tom Davidson 10 Feb 2020 21:56 UTC

  2 points

  in reply to: Wei Dai’s comment on: What can the prin­ci­pal-agent liter­a­ture tell us about AI risk?

  Re the difference between Monopoly rents and agency rents: monopoly rents would be eliminated by competition between firms whereas agency rents would be eliminated by competition between workers. So they’re different in that sense.