Dušan D. Nešić (Dushan)

PIBBSS Fel­low­ship 2025: Boun­ties and Co­op­er­a­tive AI Track Announcement

Ap­ply to the 2025 PIBBSS Sum­mer Re­search Fellowship

Ret­ro­spec­tive: PIBBSS Fel­low­ship 2024

• Dušan D. Nešić (Dushan) 14 Nov 2024 9:32 UTC

  1 point

  0 ∶ 0

  in reply to: Chris Leong’s comment on: When is Philos­o­phy Worth Fund­ing?

  In PIBBSS, we’ve had a mentor note that for alignment to go well, we need more philosophers working on foundational issues in AI rather than more prosaic researchers. I found that interesting, and I currently believe that this is true. Even in short-timeline worlds, we need to figure out some philosophy FAST.

• Dušan D. Nešić (Dushan) 13 Nov 2024 15:45 UTC

  3 points

  0 ∶ 0

  on: How would your pro­ject use ex­tra fund­ing? (Marginal Fund­ing Week 2024)

  Answering on behalf of PIBBSS, as ED of Operations.

  We have our Manifund page, which goes in-depth here:

  PIBBSS Manifund

  In brief, PIBBSS is an AI Safety org that does both field-building and research, mostly focused in non-prosaic directions. We organize (in ascending order of seniority of attendees and cost) reading groups, summer fellowships, horizon scanning, and research affiliate programs.

  Marginally, funds that you donate would most likely go to either affiliate salaries or fellowship costs (~20.000 USD per marginal fellow for three months based on last year’s costs). This requires us to get funding for our fixed costs, but we are fairly optimistic, although our runway expires in ~6 months. If you are a large funder, please get in touch, as we have the capacity for ~3 million or more in field-building and research over the next 18 months.

  Reach out with any questions!

  What links here?

  • Meet the can­di­dates in the Fo­rum’s Dona­tion Elec­tion (2024) by Toby Tremlett🔹 (18 Nov 2024 12:10 UTC; 75 points)

• Dušan D. Nešić (Dushan) 16 Oct 2024 18:54 UTC

  1 point

  0 ∶ 0

  in reply to: Arepo’s comment on: Are there stand­out EA digi­tal no­mad hubs?

  ACs are ubiquitous and our office has two, we’re not like EU countries ;)

• Dušan D. Nešić (Dushan) 14 Oct 2024 19:10 UTC

  8 points

  0 ∶ 0

  in reply to: Arepo’s comment on: Are there stand­out EA digi­tal no­mad hubs?

  Sorry, was traveling and only saw this now (thanks Gergo)! Generally many places are visa free to us, and if you have Schengen, UK or US visa valid (and have actually traveled to those places at least once on that visa) you can come visa free to Serbia. Visa is also generally good to get, but embassies don’t exist in all countries so sometimes you have to ship your passport (from Philippines to Indonesia for example), so do plan accordingly. Happy to help write an invitation letter, but more important for the visa is having booked accommodation, there’s cheap hostels walking distance from the office, I’m happy to send links.

• Dušan D. Nešić (Dushan) 2 Oct 2024 12:31 UTC

  1 point

  0 ∶ 0

  in reply to: Joseph’s comment on: Reflec­tions on the sec­ond EA Ser­bia/​Hun­gary Subot­ica Retreat

  Yes! The main reason for a relatively detailed write up is because I think this model can be easily replicated and is very much not dependant on our region.

• Dušan D. Nešić (Dushan) 25 Sep 2024 16:14 UTC

  6 points

  0 ∶ 0

  in reply to: calebp’s comment on: Reflec­tions on the sec­ond EA Ser­bia/​Hun­gary Subot­ica Retreat

  Thanks for your thoughts, Caleb! Yes, I come from 10 years in orgs where most retreats are self-funded, and we always had a good time—learning that in EA retreats are funded was a culture shock for me. For bigger conferences, I still think it is important to be able to give free tickets and even food and travel subsidies in part or full because diversity the of thought and inclusion of those who cannot afford it matters a lot. For something this small, where I expect we can recognize such cases, I think self-funded events are quite great.

Reflec­tions on the sec­ond EA Ser­bia/​Hun­gary Subot­ica Retreat

An­nounc­ing the PIBBSS Sym­po­sium ’24!

• Dušan D. Nešić (Dushan) 17 Aug 2024 7:59 UTC

  6 points

  2 ∶ 0

  on: This chart is right. Most in­ter­ven­tions don’t do much. (Cameroon ex­pe­rience)

  Thank you for this amazing write up. I had suspected that interventions at the level of charity would not work, compared to interventions at the level of funders. If funders require effectiveness, charities will care—but except some very newly started orgs, I’ve felt that institutional inertia stops them from caring about the mission and instead they just Goodheart the things funders want or institution has focused on. This makes me update even more away from Local Priorities Research, and towards Contextualisation work even in countries where there must be some cost effective intervention to do. Perhaps a new cause area of “Funder Sophistication” or something is needed to make top-down changes in attitude? Anyhow, thank you for your work and for this write-up!

• Dušan D. Nešić (Dushan) 16 May 2024 22:17 UTC

  3 points

  0 ∶ 0

  in reply to: trammell’s comment on: Notes on risk compensation

  I am saying something like: If actual risk is reduced by a quantity A, but the perceived reduction is actually B, then it’s worth spending time telling people the difference in cases where B>>A, else the effects of the project have been negative (like the effects of green-washing or more relevantly safety-washing). This is not about AI safety in general but for a particular intervention on a case by case basis.

• Dušan D. Nešić (Dushan) 15 May 2024 13:42 UTC

  1 point

  0 ∶ 0

  on: Notes on risk compensation

  A corollary of this might be “it is worthwhile to put effort into showing how little risk mitigation has actually been accomplished by the risk-reducing efforts” in order to minimize risk compensation.

• Dušan D. Nešić (Dushan) 10 Apr 2024 15:01 UTC

  7 points

  1 ∶ 0

  on: Should EA ‘com­mu­ni­ties’ be ‘pro­fes­sional as­so­ci­a­tions’?

  I think I broadly agree. It would also help with attracting actual professionals. The “bar to entry” decision is talking price, but I am generally interested in it. The thing we lose is the ability to say is “Well, that person wasn’t REALLY EA” but the thing we gain is the ability to say is “Well, that person was kicked out of EA/​never joined” or something like that when talking about a bad actor who has not acted in an EA way but has connections to the community. I don’t think that should be our main consideration, but it is worthwhile to know that the lever exists to distance ourselves from bad actors. My main consideration here is that it does look more serious, I’ve written before about Rotary and AIESEC and their membership, and how being a member has obligations and is a much clearer thing than EA movement as is.

  On the other hand, I suppose in universities where EA clubs are university clubs they do have that more formal form, so perhaps we can investigate how things look like there compared to geographical communities, see if anything noticeable is different compared to countries where the clubs are not formal university clubs (I believe many countries do not have the western concept of University club, I think EA Hungary had student clubs not incorporated in universities for example). Probably there’s too many other factors also correlated which confounds the study, but maybe we’d have enough data points?

• Dušan D. Nešić (Dushan) 5 Apr 2024 11:12 UTC

  1 point

  0 ∶ 0

  in reply to: Dicentra’s comment on: In­tro­duc­ing “Bribe Well”

  Fantastic, I didn’t know about this, I would have included it in the post! :D

• Dušan D. Nešić (Dushan) 5 Apr 2024 11:11 UTC

  1 point

  1 ∶ 0

  in reply to: Ian Turner’s comment on: In­tro­duc­ing “Bribe Well”

  There’s something about the fact that having dissolved justice system in order to rule, dictators are often afraid of stepping down, since there’s no legal process to protect them once they are no longer president. Thus, a credible island paradise where all countries in the world agree to send leaders who decide to peacefully retire would be a good EV; including good or mediocre rulers, so that there’s no incentive to be extra bad. Maybe it brings in bad motivation to rule in the first place, but I am sure we could figure it out by next April 1st and turn it into a post :D

• Dušan D. Nešić (Dushan) 1 Apr 2024 18:37 UTC

  7 points

  0 ∶ 0

  in reply to: NickLaing’s comment on: In­tro­duc­ing “Bribe Well”

  Yeah, ChatGPT—this is first response, few edits, and it was unusually full of zingers. Maybe the bullet points I used mentioning Eastern Europeans brought our sense of humor over to it.

• Dušan D. Nešić (Dushan) 1 Apr 2024 18:27 UTC

  6 points

  0 ∶ 0

  in reply to: Jason’s comment on: In­tro­duc­ing “Bribe Well”

  Oh, another stellar contribution! I mean, if their reluctance to bribes generalizes to our adversaries, then that’s some positive externallity; but the downside risk of adversaries getting to do all the “Bribin’ ” (as we call it in the biz) and us doing none is too high. Still, “if none of your prospects ended up in jail, your “funding” bar was too high”, as the saying goes! Maybe we just do BribeWell, and simultaneously run a secret market for how much money we could get paid if we revealed our transactions, allowing us to “cash-out” when we see a high enough impact opportunity (bribing AGI for example).

  New Alignment Agenda—Make AI Bribable.

• Dušan D. Nešić (Dushan) 1 Apr 2024 18:25 UTC

  8 points

  0 ∶ 0

  in reply to: NickLaing’s comment on: In­tro­duc­ing “Bribe Well”

  I shamelessly admit that AI wrote all of those, leaving me in awe. I thought I’d need to write in some zingers, but honestly, it was just my job to get out of its way. There’s a metaphor there somewhere...