Peter Slattery 🔸

New role: Se­nior Re­searcher—MIT AI Risk Initiative

[New Paper] Pri­ori­tiz­ing Risks from AI: A Delphi Study of 272 Experts

• Peter Slattery 🔸 18 May 2026 16:00 UTC

  15 points

  5 ∶ 0

  on: AI safety is ex­tremely bot­tle­necked on grantmakers

  Thank you for this. I agree that there is an urgent need for more grantmakers, based on my own experience and on conversations with people who are actively involved in the space.

  I have thought about this a little and discussed it with some grantmakers at Coefficient Giving, so I wanted to briefly mention a few ideas in case they have not already been considered.

  First, have you considered headhunting strong forecasters or superforecasters? My intuition is that demonstrated forecasting ability may be a useful signal of grantmaking potential.

  Second, have you considered making grants specifically aimed at improving the grantmaker pipeline? I think there are many possible projects and programmes that could increase the supply of capable grantmakers, and it would be very valuable to support more work in this area, both for Coefficient Giving and for the community as a whole.

  Third, one specific pipeline project I would encourage you to consider is a more tiered grantmaking ecosystem. I suspect that many people would be willing and able to start by making microgrants, similar to BlueDot Rapid Grants, or by doing part-time grantmaking, similar to SFF. Their performance could then be evaluated, and the most promising people could gradually be given more responsibility and a larger grantmaking budget.

  For example, I could imagine a Coefficient microgranting programme that is open to a wider range of people and explicitly presented as a potential pathway into a full-time grantmaking role. Another possibility would be for Coefficient Giving to direct more near-miss applicants toward opportunities where they can gain relevant grantmaking experience, while also providing more support to related organisations that could help develop this talent.

  My impression is that the bar for a grantmaking role at Coefficient Giving is currently so high that many promising people may not be willing or able to spend their scarce time and motivation on an intense application process, especially if the odds of success seem low and they are not yet sure whether they would be good at the work. A more gradual pathway could help reduce these barriers while also giving Coefficient Giving better evidence about who is likely to perform well.

• Peter Slattery 🔸 11 May 2026 19:29 UTC

  2 points

  0 ∶ 0

  in reply to: Peter Slattery 🔸’s comment on: The ta­bles have turned on AI sceptics

  Oh, and the preprint will almost certainly be submitted for peer review, but it might take 1-2 years before it is published.

• Peter Slattery 🔸 11 May 2026 19:28 UTC

  6 points

  1 ∶ 0

  in reply to: Yarrow Bouchard 🔸’s comment on: The ta­bles have turned on AI sceptics

  
  Are you sure you are thinking of the correct organization when you say:

  FutureTech seems to be one of those weird MIT hybrids between an academic research group and a management consultancy. I’m not sure if they’ve ever published a peer-reviewed paper.

  I say that because the lab has many publications, including in top peer-reviewed journals like Science. For more context, here is the publications page and here is the bio for Neil Thompson, the head of the lab:
  
  Dr. Thompson’s work has over 3000 citations with an h-index of 21 across his publication portfolio, including such well known and renowned papers as Expertise, The Computational Limits of Deep Learning, and There’s plenty of room at the Top: What will drive computer performance after Moore’s law? Dr. Thompson has been invited to present his work and recommendations to Congressional Staffers (House and Senate), the US Federal Reserve, the Pentagon, National Security Staff, the Department of Commerce, the Department of Energy, Brookings Institute, and most recently presented at a World Summit on the same program as the Prime Minister of India and Former Prime Ministers of England and Australia. With experience in 80+ countries, Dr. Thompson’s research and impact is on a global scale.

• Peter Slattery 🔸 12 Apr 2026 23:25 UTC

  2 points

  0 ∶ 0

  on: What it’s like to be an AI safety grant­maker (and why we need more of them)

  Thanks for this, I entirely agree.
  
  I’d love to see more infrastructure put in place to resolve this problem. I’d like current funders to aim for a world where we have so many capable grantmakers that some are occasionally underused because there is not enough funding to allocate.
  
  For instance, that might involve programs to proactively identify people with skills relevant to grantmaking, such as forecasting, headhunting, and transparent reasoning, funding them generously to participate in grantmaking fellowships, and placing them in grantmaking organizations when there is a strong fit.

Se­nior Re­searcher—MIT AI Risk Initiative

ML Eng­ineer—MIT AI Risk Ini­ti­a­tive, Con­trac­tor, Part-time, 6-months

• Peter Slattery 🔸 1 Jan 2026 13:56 UTC

  2 points

  0 ∶ 0

  in reply to: Jeff Kaufman 🔸’s comment on: Don’t Sell Stock to Donate

  My understanding is that I can simply need to print and send the letter/​email from Every.org to Interactive Brokers, and they will facilitate the transfer.

  One thing to note here is that end-of-year transfers can only be facilitated if the request has been made by the start of December. Another is that because I submitted my request too late in December, I will have to submit it again in January.
  
  However, all things considered, if they do facilitate the transaction based on the submission of the email/​letter shared from Every.org it will feel like a relatively straightforward process for me. However, maybe I’m going to find out later that there are some other complications in the process that I’ve missed?

• Peter Slattery 🔸 27 Dec 2025 23:28 UTC

  3 points

  0 ∶ 0

  in reply to: Alexander Saeri’s comment on: EAGxAus­trala­sia 2025 Ret­ro­spec­tive—It went well and we learnt things that might be use­ful for you

  I agree!

• Peter Slattery 🔸 26 Dec 2025 13:08 UTC

  2 points

  0 ∶ 0

  in reply to: Lorenzo Buonanno🔸’s comment on: Why aren’t more EA or­ga­ni­za­tions us­ing Every.org, or some­thing similar, to ac­cept stock and crypto dona­tions?

  Re: Googling, this post wasn’t really motivated by finding out how to donate crypto/​stocks to AMF. I mainly just wanted to signal to EA orgs that they should be able to receive such donations (or understand why some might not accept it).

  The rough sequence of events was:

  • I started by trying to donate stock/​crypto to Horizon and didn’t see an obvious pathway on their site.

  • I found their (inactive) profile on Every.org and noticed that a number of other EA-aligned orgs weren’t (active) there either.

  • I searched things like “how to donate cryptocurrency to Effective Altruism charities” and ran into posts/​resources that seemed outdated.

  • Then I did a quick (and clearly insufficient) scan of GWWC and AMF and mistakenly concluded they didn’t support these options.

  • I then felt that there was some reasonable probability that there might be a blind spot/​gap here and posted the question.

  Given my experience, I wonder if it would be helpful for someone (CEA?) to maintain a simple resource listing which EA orgs accept stock and crypto? I’m not sure if there is enough confusion or demand to justify it, though.

• Peter Slattery 🔸 26 Dec 2025 1:30 UTC

  2 points

  0 ∶ 0

  in reply to: Lorenzo Buonanno🔸’s comment on: Why aren’t more EA or­ga­ni­za­tions us­ing Every.org, or some­thing similar, to ac­cept stock and crypto dona­tions?

  Thanks for correcting me, Lorenzo, I really appreciate it! I was clearly a little inattentive/​hasty when doing my research and writing this post. I have added a note to the original and will downvote myself for spreading false info, and wasting people’s time here.

[Question] Why aren’t more EA or­ga­ni­za­tions us­ing Every.org, or some­thing similar, to ac­cept stock and crypto dona­tions?

• Peter Slattery 🔸 7 Nov 2025 14:27 UTC

  2 points

  0 ∶ 0

  in reply to: Christoph Hartmann 🔸’s comment on: You should prob­a­bly track your time (and it just got eas­ier)

  Yeah, I can see how that’s a bit of a dilemma. To avoid the flags, maybe you could make it a separate download or something people have to specifically email you to request.

  I can also imagine that for some people, especially those in certain jobs, the screenshots or logging features won’t be worth the risk. On that note, I decided to uninstall the tool myself, given the security considerations and the requirements from my employer.

• Peter Slattery 🔸 4 Nov 2025 16:47 UTC

  2 points

  0 ∶ 0

  in reply to: Peter Slattery 🔸’s comment on: You should prob­a­bly track your time (and it just got eas­ier)

  Ok, I have now read some comments where you explain the privacy below, and watched your video. I’m going to engage with the university security team and ask them what they think, and whether they feel this is appropriate to have on a work computer.

• Peter Slattery 🔸 4 Nov 2025 16:44 UTC

  4 points

  0 ∶ 0

  on: You should prob­a­bly track your time (and it just got eas­ier)

  I looked into this and no longer have these concerns.
  
  Thank you for this. I have some security concerns, and I am uninstalling the software after I received the message below from my university. Can you please respond here to explain?

  “IS&T Information Security received a Crowdstrike alert about some software you downloaded from donethat.ai. Presumably it was intentional, but this software comes installed with what looks like keylogger functionality. …Having a keylogger on your system is pretty risky. Even if they purport to not store data, there is always the risk that the software or your machine is compromised, and the data stream is intercepted for misuse”

  ...

  This installer wrote the following file, which is associated with keylogging functionality:

  FILE: …\node_modules\node-global-key-listener\bin\WinKeyServer.exe”
  

MIT Fu­tureTech are hiring a Post­doc­toral As­so­ci­ate to work on AI Perfor­mance and Safety

• Peter Slattery 🔸 16 May 2025 11:17 UTC

  2 points

  0 ∶ 0

  on: AI for An­i­mals 2025 Bay Area Retrospective

  Well done!

Cam­bridge Bos­ton Align­ment Ini­ti­a­tive Sum­mer Re­search Fel­low­ship in AI Safety (Dead­line: May 18)

• Peter Slattery 🔸 11 May 2025 15:30 UTC

  2 points

  0 ∶ 0

  in reply to: Jason’s comment on: EA Funded Kick­start­ing For Good in­cu­ba­tor seek­ing ap­pli­ca­tions (dead­line 23 May 2025)

  Sorry, thanks for pointing that out!