zeshen

• zeshen 14 Mar 2024 12:36 UTC

  1 point

  0 ∶ 0

  in reply to: Lorenzo Buonanno🔸’s comment on: What posts would you like some­one to write?

  Thanks! I think I was having the impression that the Gates Foundation was struggling to give out money (e.g. this comment from a long time ago), but I’m now learning that that’s probably no longer true—they set a goal of $9 billion by 2026 and they’re already having a budget of $8.6 billion this year. Now it makes sense.

• zeshen 13 Mar 2024 18:57 UTC

  3 points

  0 ∶ 0

  in reply to: DavidNash’s comment on: What posts would you like some­one to write?

  Thanks for the link! I vaguely remember reading this but probably didn’t really get an answer that I was hoping for. In the case of AMF, reason 1 doesn’t apply, because they seem to want the money to do things now instead of building reserves. Reason 4 seems most relevant—maybe the Gates Foundation is hoping that a Malaria vaccine (which recent developments have shown positive results) could render bed nets futile? But I don’t think I buy this either—considering how effective these vaccines currently are, how long it takes to roll out vaccines in these countries, and that Bill Gates himself has previously vouched for bed nets (albeit before the vaccines were endorsed by WHO). As for reasons 2, 3, and 5, I just don’t really see how these reasons are worth killing so many babies for—I can’t picture a decision maker in the Foundation saying “yeah we have decided to let a hundred thousand people die of Malaria so that we can diversify our risks and encourage others to donate”.

  I may be missing something, but I only see a few reasonable scenarios:

  1. The Gates Foundation does indeed plan to donate, and they might be the ‘donor of last resort’

  2. They really do not intend to fill the funding gap, perhaps because they don’t think additional funding to AMF is as cost-effective as advertised

  3. They are confident that AMF will somehow get funding from other sources

• zeshen 12 Mar 2024 7:10 UTC

  6 points

  1 ∶ 0

  on: What posts would you like some­one to write?

  I’d be interested to understand why there are still huge shortfalls in the supposedly top effective charities.

  For example, AMF has a funding gap of $300 milion. The Bill and Melinda Gates Foundation has an endowment of $67 billion, which of course they intend to donate away. Bill Gates also endorses GiveWell, and has an explicit focus on solving Malaria (it also lists 20 organizations that they partner with, but AMF is not one of it).

  So why isn’t the AMF funding gap plugged yet, by the Gates foundation, or anyone else? As for the Foundation, is it a matter of grant evaluation process? Is there anything else relevant I should know to better understand the whole funding landscape of these issues?

• zeshen 16 Apr 2023 7:10 UTC

  1 point

  0 ∶ 0

  on: AGI Isn’t Close—Fu­ture Fund Wor­ld­view Prize

  To achieve AGI we will need to program the following:

  • knowledge creating processes

  • emotions

  • creativity

  • free will

  • consciousness.

  I suspect a large part of the crux is the definition of AGI itself. I don’t know many people who think that an agent /​ system must fulfill all of the above criteria to qualify as ‘AGI’. I personally use the term AGI to refer to systems that have at least human-level capabilities at all tasks that a human is capable of performing, regardless of whether the system posses other properties like consciousness and free will.
  
  On a separate matter, I think it might be a good idea if there is a dual voting system for posts, just like comments, where people can upvote/​downvote and also agree/​disagree vote. This is a post that I would upvote but strong disagree on. In the meantime I gave it an upvote anyway, since I like posts that at least attempt to constructively challenge prevailing worldviews, and also to balance out all the downvotes.

• zeshen 25 Mar 2023 6:00 UTC

  4 points

  0 ∶ 0

  in reply to: gabriel_wagner’s comment on: Sur­vey on in­ter­me­di­ate goals in AI governance

  I just browsed through it, their reasons for not doing so is also described in a section in the report.

• zeshen 18 Mar 2023 3:13 UTC

  1 point

  0 ∶ 0

  on: Ev­i­dence on how cash trans­fers em­power women in poverty

  I think it’ll be great if this post is also on GiveDirectly’s website, perhaps under the blogs section?

• zeshen 18 Jan 2023 6:36 UTC

  2 points

  0 ∶ 0

  on: aisafety.com­mu­nity—A liv­ing doc­u­ment of AI safety communities

  Crossposting plex’s comment from LessWrong:

  Updates!

  1. Have been continually adding to this, up to 49 online communities.

  2. Added a student groups section, with 12 entries and some extra fields (website, contact, calendar, mailing list), based on the AGISF list (in talks with the maintainer to set up painless syncing between the two systems).

  3. Made a form where you can easily add entries.

  Still getting consistent traffic, happy to see it getting used :)

• zeshen 16 Jan 2023 0:49 UTC

  1 point

  0 ∶ 0

  in reply to: firstuserhere’s comment on: aisafety.com­mu­nity—A liv­ing doc­u­ment of AI safety communities

  Let me know if it still doesn’t work, thanks!

• zeshen 15 Jan 2023 9:04 UTC

  2 points

  0 ∶ 0

  on: Con­cerns about AI safety ca­reer change

  Great questions.
  On question 4, I don’t personally know of any groups based in Asia, but feel free to check out this database of AI Safety relevant communities, and join any of them.

• zeshen 8 Jan 2023 8:44 UTC

  1 point

  0 ∶ 0

  in reply to: NunoSempere’s comment on: Read The Sequences

  Would you recommend Probability Theory: The Logic of Science to people with little math background?

• zeshen 1 Dec 2022 1:26 UTC

  1 point

  0 ∶ 0

  on: Why Neu­ron Counts Shouldn’t Be Used as Prox­ies for Mo­ral Weight

  This is nice, but I’d also be interested to see quantification of moral weights for different animals when accounting for all the factors besides neuron count, and how much it differs from solely using neuron count alone.

• zeshen 16 Nov 2022 7:42 UTC

  11 points

  3 ∶ 1

  on: Who’s at fault for FTX’s wrongdoing

  I like this post much more than your previous post.

A new­comer’s guide to the tech­ni­cal AI safety field

aisafety.com­mu­nity—A liv­ing doc­u­ment of AI safety communities

• zeshen 11 Oct 2022 8:53 UTC

  1 point

  0 ∶ 0

  in reply to: Emrik’s comment on: Why defen­sive writ­ing is bad for com­mu­nity epistemics

  Confidence-in-path rather than confidence-in-results.

  Nicely said.

  --

  See you at GatherTown soon!

• zeshen 10 Oct 2022 23:16 UTC

  8 points

  3 ∶ 0

  on: Why defen­sive writ­ing is bad for com­mu­nity epistemics

  I really like this post, especially as someone who is fairly anxious when writing for fear of being judged as ignorant. I definitely agree that we should promote an environment conducive for people to say wrong things.

  However, I don’t fully agree with the notion of celebrating the self-confidence of people who “declare that think they can do more good than Peter Singer”. I’m quite likely misinterpreting what you mean by confidence as over-confidence, but just on the face value of it, I prefer claims to have an appropriate level of confidence associated with them. When someone makes a strong claim, I’d like to know whether the person is a domain expert with good epistemics who has done extensive research to arrive at the conclusion, or an innocent kid who claims to have discovered the most important thing in the world. Perhaps just stating epistemic status upfront would solve the problem. And perhaps over-confident people who are just about to take your advice to celebrate confidence and stop rewarding modesty should just reverse the advice.

  On a tangential note, I sometimes find myself doing ‘defensive writing’ not just for defensive reasons, but also to try to convey what I mean to the reader as accurately as I can by ruling out everything else.

• zeshen 7 Oct 2022 12:40 UTC

  5 points

  0 ∶ 0

  in reply to: Stephen Clare’s comment on: Ask (Every­one) Any­thing — “EA 101”

  Looking at your profile I think you have a good idea of answers already, but for the benefit of everyone else who upvoted this question looking for an answer, here’s my take:

  Are there AI risk scenarios which involve narrow AIs?

  Yes, a notable one being military AI i.e. autonomous weapons (there are plenty of related posts on the EA forum). There are also multipolar failure modes on risks from multiple AI-enabled superpowers instead of a single superintelligent AGI.

  Why does most AI risk research and writing focus on artificial general intelligence?

  A misaligned AGI is a very direct pathway to x-risk, where an AGI that pursues some goal in an extremely powerful way without having any notion of human values could easily lead to human extinction. The question is how to make an AI that’s more powerful than us do what we want it to do. Many other failures modes like bad actors using tool (narrow) AIs seem less likely to lead directly to x-risk, and is also more of a coordination problem than a technical problem.

• zeshen 24 Jul 2022 6:47 UTC

  14 points

  0 ∶ 0

  in reply to: titotal’s comment on: How EA is per­ceived is cru­cial to its fu­ture trajectory

  Point taken, though EA seems to welcome criticisms a lot to the point of giving out large prizes for good criticisms. Scott Alexander also argues that EA actually goes too far in taking criticisms on board.
  
  

• zeshen 29 Apr 2022 22:59 UTC

  4 points

  0 ∶ 0

  on: Com­par­a­tive ad­van­tage does not mean do­ing the thing you’re best at

  Thanks for the article! I was a little confused by the title “doing the thing you’re best at” on whether it means “doing the thing you’re best at compared to everything else you know how to do” or “doing the thing you’re best at compared to everyone else”, where the former is comparative advantage and the latter is absolute advantage? The content of your article is clearer though—thanks for that.

• zeshen 27 Apr 2022 3:17 UTC

  2 points

  0 ∶ 0

  on: [$20K In Prizes] AI Safety Ar­gu­ments Competition

  “Anyone who looked for a source of power in the transformation of the atoms was talking moonshine.”—Ernest Rutherford, shortly before the invention of the Atomic Bomb.
  
  ”If there were any risk of [an “AI apocalypse”], it wouldn’t be for another few decades in the future.”—Yann LeCun, 2016 (not being really serious)