On the way out of EA.
đ¸ 10% Pledger.
Likes pluralist conceptions of the good.
Dislikes Bay Culture being in control of the future.
JWS đ¸
Karma: 4,880
Full disclosureâI read a draft of this piece and provided titotal with some feedback on it
Some high level reflections on this piece and interpreting the surrounding Debate:
I think this is an excellent piece, and I think well worth reading, especially for EAs who are tempted to easily defer to âhigh statusâ organisations/âindividuals in the field of AI Safety. Having said that, I also think @elifland and @kokotajlod deserve a lot of credit for being open to criticism and working with titotal privately, and communicating in good faith here, on LW, titotalâs substack etc[1]
I think the piece shows that there are clear flaws with how the model is constructed, and that its design betrays the assumptions of those creating it. Of course this is a feature, not a bug, as itâs meant to be a formalisation of the beliefs of the AI2027 team (at least as I understand it). This is completely ok and even a useful exercise, but then I think nostalgebraist is accurate in saying that if you didnât buy the priors/âarguments for a near-term intelligence explosion before the model, then you wonât afterwards. The arguments for our assumptions, and how to interpret the data we have are ~the whole ball game, and not the quantitative forecasts/âscenarios that they then produce.
I particularly want to draw attention to the âSix stories that fit the dataâ section, because to my mind it demonstrates the core issue/âepistemological crux. The whole field of AI Safety is an intense case of the underdetermination of theory given evidence,[2] and choosing which explanation of the world to go with given our limited experience of it is the key question of epistemology. But as titotal points out, a similar exercise to AI2027 could pick any one of those curves (or infinitely many alternatives) - the key points here are the arguments and assumptions underlying the models and how they clash.
Why the AI debates seem so intractable is a combination of:
We have limited data with which to constrain hypotheses (METRâs curve has 11 data points!)
The assumptions that underly the differences are based on pretty fundamental worldview differences and/âor non-technical beliefs about how the world worksâsee AI as Normal Technology[3] or Ajeya Cotraâs appearance on the AI Summer Podcast
The lack of communication between the various âdifferent campsâ involved in frontier AI and AI research contributes to misunderstandings, confrontational/âadversarial stances etc.
The increased capability and saliency of AI in the world starting to lead to political polarisation effects which might make the above worse
Linked to 4, what happens with AI seems to be very high stakes. Itâs not just opinions differ, but the range of what could happen is massive. Thereâs a lot at risk if actions are taken which are later proben to be misguided.
Given this epistemological backdrop, and the downward spiral in AI discourse over the last 2 years,[4] I donât know how to improve the current state of affairs apart from âlet reality adjudicate the winnersâ - which often leaves me frustrated and demotivated. Iâm thinking perhaps of adversarial collaborations between different camps, boosting collaborative AI Safety strategies, and using AI to help develop high-trust institutions.[5] But I donât think exercise like AI2027 push the field forward because of the âpeak forecastingâ involved, but instead by surfacing the arguments which underline the forecasts for scrutiny and falsification.
- ^
Though I think Alfredo Parraâs recent post is also worth bearing in mind
- ^
Though, tbf, all of human knowledge is. This problem isnât unique to AI/âAI Safety
- ^
Particularly âThe challenge of policy making under uncertaintyâ section
- ^
A spiral of which no side is blameless
- ^
If youâre interested in collaborating on/âsupporting any of the above, please reach out
I was going to reply with something longer here, but I think Gregory Lewisâ excellent comment highlights most of what I wanted to, r.e. titotal does actually give an alternative suggestion in the piece.
So instead Iâll counter two claims I think you make (or imply) in your comments here:
1. A shoddy toy model is better than no model at all
I mean this seems clearly not true, if we take model to be referring to the sort of formalised, quantified exercise similar to AI-2027. Some examples here might be Samuelsonâs infamous predictions of the Soviet Union inevitably overtaking the US in GNP.[1] This was a bad model of the world, and even if it was âbetterâ than the available alternatives or came from a more prestigious source, it was still bad and I think worse than no model (again, defined as formal exercise ala AI2027).
A second example I can think of is the infamous Growth in a Time of Debt paper, which I remember being used to win arguments and justify austerity across Europe in the 2010s, being rendered much less convincing after an Excel error was corrected.[2]
TL;dr, as Thane said on LessWrong, we shouldnât grade models on a curve
2. You need to base life decisions on a toy model
This also seems clearly false, unless weâre stretching âmodelâ to mean simply âa reason/âargument/âjustificationâ or defining âlife decisionsâ narrowly as only those with enormous consequences instead of any âdecision about my lifeâ.
Even in the more serious cases, the role of models is to support presenting arguments for or against some decision or not, or to frame some explanation about the world, and of course simplification and quantification can be useful and powerful, but they shouldnât be the only game in town. Other schools of thought are available.[3]
- ^
- ^
The reproduction paper turned critique is here, feels crazy that I canât see the original data but the âmodelâ here seemed just to be spreadsheet of ~20 countries where the average only counted 15
- ^
Such as:
Make a decision based on the best explanation of the world
Go with common-sense heuristics since they likely encode knowledge gained from cultural evolution
- ^
As with many âp->qâ arguments, I think this one is true for the trivial reason that q holds independent of p. I.e. itâs true that itâs unlikely that people will identify as EAâs one hundred years from now[1] but that also was unlikely 5 years ago.
This is a good and valid point for sure. I suppose that for the 4 failed movements I bring up in the post, they all failed to achieve their goals in their own terms and their ideas failed to influence other movements.[1] I think the q of âEffective Altruism ends as a movementâ is likely because the rate of dying out for all movements is 100%, just like it is for all living beings
So perhaps I want to distinguish between:
1) Movements that âdie outâ because they succeed enough that their ideas permeate into the mainstream and outlive the initial social/âintellectual movement
2) Movements that âdie outâ because they lose enough reputation and support that nobody carries those ideas forward.This would be different as well from:
3) The Movementâs goals eventually being realised
4) The Movement being a force for good in the world
So The Chartists might be an example of 2 & 3 - after the 1848 demostration they basically completely faded from power, but by 1918 5 out of 6 Chartist reforms had been implemented. Itâs not clear to what extent the movement was causally responsible for this though
I think Revolutionary Marxism of various forms might be 1 & 4 - it was hugely popular in the late 19th and early 20th century even after Marx died, or the waning of power of explicitly Marxist parties, the ideas still had massive influence and can be casually traced to those intellectuals I think. I nevertheless think that their influence has been very negative for the world, but YMMV[2]
So I guess the underlying question is, if EA is in a prolonged or terminal decline (as in, we expect no early-Quaker style reforms to arrest the momentum, which is not guaranteed) then is it of form 1 or 2? Iâm not sure, itâs an open question. I think conditioning on an SBF-scale reputational damage and subsequent âevaporate coolingâ of the movement since, the odds should have moved toward 2, but itâs not guaranteed for sure and itâd be interesting to see examples of which social movements match 1 vs 2.
- ^
Thereâs maybe some wiggle room around New Atheism/âTechnocracy, but the case is harder to make if you think theyâre causally responsible
- ^
I donât want to get into a huge debate about Marxism or not, itâs just the first thing that came to mind. If you are, you could just substitute âRevolutionary Marxismâ for âNeoliberal Capitalismâ of Hayek et al in the 20th century, which had a massively successful impact on the Reagan and Thatcher administrations, for instance
- ^
Iâm going to actually disagree with your initial premiseâthe basic points are that the expected number of people in the future is much lower than longtermists estimateâbecause, at least in the Reflective Altruism blog series, I donât see that as being the main objection David has to (Strong) Longtermism. Instead, I think he instead argues That the interventions Longtermists support require additional hypotheses (the time of perils) which are probably false and that the empirical evidence longtermists give for their existential pessimism are often non-robust on further inspection.[1] Of course my understanding is not complete, David himself might frame it differently, etc etc.
One interesting result from his earlier Existential risk pessimism and the time of perils paper is that on a simple model, though he expands the results to more complex ones, people with low x-risk should be longtermists about value, and those with high x-risk estimates should be focused on the short term, which is basically the opposite of what we see happening in real life. The best way out for the longtermist, he argues, is to believe in âthe time of perils hypothesisâ. I think the main appeals to this being the case are either a) interstellar colonisation giving us existential security so weâre moral value isnât tethered to one planet,[2] or of course from b) aligned superintelligence allowing us unprecedented control over the universe and the ability to defuse any sources of existential risk. But of course, many working on Existential AI Risk are actually very pessimistic about the prospects for alignment and so, if they are longtermist,[3] why arenât they retiring from technical AI Safety and donating to AMF? More disturbingly, are longtermists just using the âtime of perilsâ belief to backwards-justify their prior beliefs that interventions in things like AI are the utilitarian-optimal interventions to be supporting? I havenât seen a good longtermist case answering these questions, which is not to say that one doesnât exist.
Furthermore, in terms of responses from EA itself, whatâs interesting is that when you look at the top uses of the Longtermism tag on the Forum, all of the top 8 were made ~3 years ago, and only 3 of the top 20 within the last 3 years. Longtermism isnât a used a lot even amongst EA any moreâthe likely result of negative responses from the broader intelligensia during the 2022 soft launch, and then the incredibly toxic result of the FTX collapse shortly after the release of WWOTF. So while I find @trammellâs comment below illuminating in some aspects about why there might be fewer responses than expected, I think sociologically it is wrong about the overarching reasonsâI think longtermism doesnât have much momentum in academic philosophical circles right now. Iâm not plugged into the GPI-Sphere though, so I could be wrong about this.
So my answer to your initial question is ânoâ if you mean âsomething big published post-Thorstad that responds directly or implicitly to him from a longtermist perspectiveâ. Furthermore, were they to do so, or to point at one already done (like The Case for Strong Longtermism) Iâd probably just reject many of the premises that give the case legs in the first place, such as that itâs reasonable to do risk-neutral-expected-value reasoning about the very long run future in the first place as a guide to moral action. Nevertheless, other objections to Longtermism I am sympathetic to are those from Eric Schwitzgebel (here, here) among others. I donât think this is Davidâs perspective though, I think he believes that the empirical warrant for the claims arenât there but that he would support longtermist policies if he believed they could be supported this way.
Iâm also somewhat disturbed by the implication that some proportion of the EA Brain-Trust, and/âor those running major EA/âAI Safety/âBiorisk organisations, are actually still committed longtermists or justify their work in longtermist terms. If so they should make sure this is known publicly and not hide it. If you think your work on AI Policy is justified on strong longtermist grounds, then Iâd love to see your the model used for that, and parameters used for the length of the time of perils, the marginal difference to x-risk the policy would make, and the evidence backing up those estimates. Like if 80k have shifted to be AI Safety focused because of longtermist philosophical commitments, then lets see those commitments! The inability of many longtermist organisations to do that is a sign of what Thorstad calls the regression to the inscrutable,[4] which is I think one of his stronger critiques.
- ^
Disagreement about future population estimates would be a special case of the latter here
- ^
In The Epistemic Challenge to Longtermism, Tarsney notes that:
More concretely, the case for longtermism seems to depend to a significant
extent on the possibility of interstellar settlement - ^
Note these considerations donât apply to you if youâre not an impartial longtermist, but then again, if many people working in this area donât count themselves as longtermists, it certainly seems like a poor sign for longtermism
- ^
Term coined in this blog post about WWOTF
- ^
A general good rule for life
- ^
(I am not a time-invariant-risk-neutral-totally-impartial-utilitarian, for instance)
- ^
No worries Ollie, thanks for the feedback :)
As I said, those bullet points were a memory of a draft so I donât have the hard data to share on hand. But when dealing with social movements itâs always going to be somewhat vibesyâdata will necessarily be observational and we canât travel back in time and run RCTs on whether SBF commits fraud or not. And the case studies do show that declines can go on for a very long time post major crisis. Itâs rare for movements to disappear overnight (The Levellers come closest of all the cases I found to that)
Fwiw I think that the general evidence does point to âEA is in declineâ broadly understood, and that should be considered the null hypothesis at this point. Iâd feel pretty gaslit if someone said EA was going swimmingly and unaffected by the tribulations of the last couple of years, perhaps less so if they think thereâs been a bounce back after an initial decline but, you know, Iâd want to see the data for that.
But as I said, itâs really not the (main) point of the post! Iâd love to add my points to a post where someone did try and do a deep dive into that question.
Hey Ollie, thanks for your feedback! It helped me understand some of the downvotes the post was getting which I was a bit confused by. I think you and perhaps others are interpreting the post as âHere are some case studies that show EA is in declineâ, but thatâs not what I was trying to write, it was more âEA is in decline, what historical cases can inform us about this?â Iâm not really arguing for âIs EA in decline?â in the post, in fact Iâm just assuming it and punting the empirical evidence for another time, since I was interested in bringing out the historical cases rather than EAs current state. So the tweets are meant to be indicative of mood/âsentiment but not load bearing proof. I do see that the rhetorical flourish in the intro might have given a misleading impression, so I will edit that to make the point of the piece more clear.
As for why, I mean, it does just seem fairly obvious to me, but social movements have fuzzy boundaries and decline doesnât have to be consistent. Nevertheless answering this question was a post I was planning on write and the evidence seemed fairly damning to meâfor instance:
Spikes in visiting the âEffective Altruismâ Wikipedia page seem to mainly be in response to negative media coverage, such as the OpenAI board fallout or a Guardian Article about Wytham Abbey. Other Metrics like Forum engagement that were growing preFTX clearly spike and decline after the scandal period.
Other organisations similar to EA considering rebounding away. Apparently CE/âAIM was considering this, and Rutger Bregman seems to be trying really hard to keep Moral Ambition free of EAs reputational orbit, which I think heâd be doing much less of if EAs prospects were more positive.
Previous community members leaving the Forum or EA in general, sometimes turning quite hostile to it. Some examples here are Habryka, NuĂąo Sempere, Nathan Young, Elizabeth from the more ârationalistâ side. Iâve noticed various people who were long term EAs, like Akash Wasil and John Halstead have deactivated their accounts. A lot of more left-wing EAs like Bob Jacobs seem to have started to move away too, or like Habiba moderate their stanc and relationship to EA.
This goes double so for leadership. I think loads kf the community felt a leadership vacuum post FTX. Dustin has deactivated his account. Holden has left OP, gone quiet, and might not longer consider himself EA? Every so often Ben Todd tweets something which I can only interpret as âtesting the waters before jumping shipâ. I donât think leadership of a thriving, growing movement acts this.
If you search âEffective Altruismâ on almost any major social media site (X, Bluesky, Reddit, etc) I suspect the general sentiment toward EA will be strongly negative and probably worse than it was preFTX and staying that way. There might be some counter evidence. Some metrics might have improved, and I know some surveys are showing mixed or positive things. I think Habrykaâs point that reputation is evaluated lazily rings true to me even if I disagree on specifics.
But again, the above is my memory of a draft, and Iâm not sure Iâll ever finish that post. I think hard data on a well formed version of the question would be good, but once again itâs not the question I was trying to get at with this post.
Some Positions In EA Leadership Should Be Elected
If we take this literally, then âsomeâ means âany at allâ, and I think that the amount of democratic input is >0 so this should be 100% yes.
Having said that, I think the bar is kind-of acting as a âhow much do you support democratising EAâ, and in that sense while I do support Cremer-style reforms, I think theyâre best introduced at a small scale to get a track record.
I wish this postâand others like itâhad more specific details when it comes to these kind of criticisms, and had a more specific statement of what they are really taking issue with, because otherwise it sort of comes across as âI wish EA paid more attention to my object-level concernsâ which approximately ~everyone believes.
If the post itâs just meant to represent your opinions thats perfectly fine, but I donât really think it changed my mind on its own merits. I also just donât like withholding private evidence, I know there are often good reasons for it, but it means I just canât give much credence to it tbqh. I know itâs a quick take but still, it left me lacking in actually evaluating.
I general this discussion reminds me a bit of my response to another criticism of EA/âActions by the EA Community, and I think what you view as âsub-optimalâ actions are instead best explained by people having:
bounded rationality and resources (e.g. You canât evaluate every cause area, and your skills might not be as useful across all possible jobs)
different values/âmoral ends (Some people may not be consequentialists at all, or prioritise existing vs possible people difference, or happiness vs suffering)
different values in terms of process (e.g. Some people are happy making big bets on non-robust evidence, others much less so)
And that, with these constraints accepted, many people are not actually acting sub-optimally given their information and values (they may later with hindsight regret their actions or admit they were wrong, of course)
Some specific counterexamples:
Being insufficiently open-minded about which areas and interventions might warrant resources/âattention, and unwarranted deference to EA canon, 80K, Open Phil, EA/ârationality thought leaders, etc. - I think this might be because of resource constraints, or because people disagree about what interventions are âunder investedâ vs âcorrectly not much invested inâ. I feel like you just disagree with 80K & OpenPhil on something big and you know, Iâd have liked you to say what that is and why you disagree with them instead of dancing around it.
Attachment to feeling certainty, comfort and assurance about the ethical and epistemic justification of our past actions, thinking, and deferenceâWhat decisions do you mean here? And what do you even mean by âourâ? Surely the decisions you are thinking of apply to a subset of EAs, not all of EA? Like this is probably the point that most needed something specific.
Trying to read between the linesâI think you think the AGI is going to be a super big deal soon, and that its political consequences might be the most important and consequently political actions might be the most important ones to take? But OpenPhil and 80K havenât been on the ball with political stuff and now youâre disillusioned? And people in other cause areas that arenât AI policy should presumably stop donating there and working their and pivot? I donât know, it doesnât feel accurate to me,[1] but like I canât get a more accurate picture because you just didnât provide any specifics for me to tune my mental model on đ¤ˇââď¸
Even having said all of this, I do think working on mitigating concentration-of-power-risks is a really promising direction for impact and wish you the best in pursuing it :)
- ^
As in, i donât think youâd endorse this as a fair or accurate description of your views
Noteâthis was written kinda quickly, so might be a bit less tactful than I would write if I had more time.
Making a quick reply here after binge listening to three Epoch-related podcasts in the last week, and I basically think my original perspective was vindicated. It was kinda interesting to see which points were repeated or phrased a different wayâwould recommend if your interested in the topic.
The initial podcast with Jaime, Ege, and Tamay. This clearly positions the Epoch brain trust as between traditional academia and the AI Safety community (AISC). tl;drâacademia has good models but doesnât take ai seriously, and AISC the opposite (from Epochâs PoV)
The âdebateâ between Matthew and Ege. This should have clued people in, because while full of good content, by the last hour/âhour and half it almost seemed to turn into âopenly mocking and laughingâ at AISC, or at least the traditional arguments. I also donât buy those arguments, but I feel like the reaction Matthew/âEge have shows that they just donât buy the root AISC claims.
The recent podcast Dwarkesh with Ege & Tamay. This is the best of the 3, but probably also best listened too after the first too, since Dwarkesh actually pushes back on quite a few claims, which means Ege & Tamay flush out their views moreâpersonal highlight was what the reference class for AI Takeover actually means.
Basically, the Mechanize cofounders donât agree at all with âAI Safety Classicâ, I am very confident that they donât buy the arguments at all, that they donât identify with the community, and somewhat confident that they donât respect the community or its intellectual output that much.
Given that their views are: a) AI will be a big deal soon (~a few decades), b) returns to AI will be very large, c) Alignment concerns/âAI risks are overrated, and d) Other people/âinstitutions arenât on the ball, then starting an AI Start-up seems to make sense.
What is interesting to note, and one I might look into in the future, is just how much these differences in expectation of AI depend on differences in worldview, rather than differences in technical understanding of ML or understanding of how the systems work on a technical level.
So why are people upset?
Maybe they thought the Epoch people were more part of the AISC than they actually were? Seems like the fault of the people believe this, not Epoch or the Mechanize founders.
Maybe people are upset that Epoch was funded by OpenPhil, and this seems to have lead to âAI accelerationâ? I think thatâs plausible, but Epoch has still produced high-quality reports and information, which OP presumably wanted them to do. But I donât think equating EA == OP, or anyone funded by OP, is a useful concept to me.
Maybe people are upset at any progress in AI capabilities. But that assumes that Mechanize will be successful in its aims, not guaranteed. It also seems to reify the concept of âcapabilitiesâ as one big thing which i donât think makes sense. Making a better Stockfish, or a better AI for FromSoft bosses does not increase x-risk, for instance.
Maybe people think that the AI Safety Classic arguments are just correct and therefore people taking actions other than it. But then many actions seem bad by this criteria all the time, so odd this would provoke such a reaction. I also donât think EA should hang its hat on âAI Safety Classicâ arguments being correct anyway.
Probably some mix of it. I personally remain not that upset because a) I didnât really class Epoch as âpart of the communityâ, b) Iâm not really sure Iâm âpart of the communityâ either and c) my views are at least somewhat similar to the Epoch set above, though maybe not as far in their direction, so Iâm not as concerned object-level either.
Iâm not sure I feel as concerned about this as others. tl;drâThey have different beliefs from Safety-concerned EAs, and their actions are a reflection of those beliefs.
It seems broadly bad that the alumni from a safety-focused AI org
Was Epoch ever a âsafety-focusedâ org? I thought they were trying to understand whatâs happening with AI, not taking a position on Safety per se.
âŚhave left to form a company which accelerates AI timelines
I think Matthew and Tamay think this is positive, since they think AI is positive. As they say, they think explosive growth can be translated into abundance. They donât think that the case for AI risk is strong, or significant, especially given the opportunity cost they see from leaving abundance on the table.
Also important to note is what Epoch boss Jaime says in this very comment thread.
As I learned more and the situation unfolded I have become more skeptical of AI Risk.
The same thing seems to be happening with me, for what itâs worth.
People seem to think that there is an âEA Orthodoxyâ on this stuff, but there either isnât as much as people think, or people who disagree with it are no longer EAs. I really donât think it makes sense to clamp down on âdoing anything to progress AIâ as being a hill for EA to die on.
Note: Iâm writing this for the audience as much as a direct response
The use of Evolution to justify this metaphor is not really justified. I think Quintin Popeâs Evolution provides no evidence for the sharp left turn (which won a prize in an OpenPhil Worldview contest) convincingly argues against it. Zvi wrote a response from the âLW Orthodoxâ camp that wasnât convincing and Quintin responds against it here.
On âInner vs Outerâ framings for misalignment is also kinda confusing and not that easy to understand when put under scrutiny. Alex Turner points this out here, and even BlueDot have a whole âCriticisms of the inner/âouter alignment breakdownâ in their intro which to me gives the game away by saying âtheyâre useful because people in the field use themâ, not because their useful as a concept itself.
Finally, a lot of these concerns revolve around the idea of their being set, fixed, âinternal goalsâ that these models have, and represent internally, but are themselves immune from change, or can hide from humans, etc. This kind of strong âGoal Realismâ is a key part of the case for âDeceptionâ style arguments, whereas I think Belrose & Pope show an alternative way to view how AIs work is âGoal Reductionismâ, in which framing the issues imagined donât seem certain any more, as AIs are better understood as having âcontextually-activated heuristicsâ rather than Terminal Goals. For more along these lines, you can read up on Shard Theory.
Iâve become a lot more convinced about these criticisms of âAlignment Classicâ by diving into them. Of course, people donât have to agree with me (or the authors), but Iâd highly encourage EAs reading the comments on this post to realise Alignment Orthodoxy is not uncontested, and is not settled, and if you see people making strong cases based on arguments and analogies that seem not solid to you, youâre probably right, and you should look to decide for yourself rather than accepting that the truth has already been found on these issues.[1]
- ^
And this goes for my comments too
- ^
Iâm glad someone wrote this up, but I actually donât see much evaluation here from you, apart from âitâs too early to sayâ, but then Zhou Enlai pointed out that you could say that about the French Revolution,[1] and I think we can probably say some things. I generally have you mapped to the âright-wing Rationalistâ subgroup Arjun,[2] so itâd be actually interested to get your opinion instead of trying to read between the lines on what you may or may not believe. I think there was a pretty strong swing in Silicon Valley /â Tech Twitter & TPOT /â Broader Rationalism towards Trump, and I think this isnât turning out well, so Iâd actually be interested to see people saying what they actually thinkâbe that âI made a huge mistakeâ, âIt was a bad gamble but Harris wouldâve been worseâ or even âThis is exactly what I wantâ
Hey Cullen, thanks for responding! So I think there are object-level and meta-level thoughts here, and I was just using Jeremy as a stand-in for the polarisation of Open Source vs AI Safety more generally.
Object LevelâI donât want to spend too long here as itâs not the direct focus of Richardâs OP. Some points:On âelite panicâ and âcounter-enlightenmentâ, heâs not directly comparing FAIR to it I think. Heâs saying that previous attempts to avoid democratisation of power in the Enlightenment tradition have had these flaws. I do agree that it is escalatory though.
I think, from Jeremyâs PoV, that centralization of power is the actual ballgame and what Frontier AI Regulation should be about. So one mention on page 31 probably isnât good enough for him. Thatâs a fine reaction to me, just as itâs fine for you and Marcus to disagree on the relative costs/âbenefits and write the FAIR paper the way you did.
On the actual points though, I actually went back and skim-listened to the the webinar on the paper in July 2023, which Jeremy (and you!) participated in, and man I am so much more receptive and sympathetic to his position now than I was back then, and I donât really find Marcus and you to be that convincing in rebuttal, but as I say I only did a quick skim listen so I hold that opinion very lightly.
Meta Level -
On the âescalationâ in the blog post, maybe his mind has hardened over the year? Thereâs probably a difference between ~July23-Jeremy and ~Nov23Jeremy, which he may view as an escalation from the AI Safety Side to double down on these kind of legislative proposals? While itâs before SB1047, I see Wiener had introduced an earlier intent bill in September 2023.
I agree that âpeople are mad at us, weâre doing something wrongâ isnât a guaranteed logic proof, but as you say itâs a good prompt to think âshould i have done something different?â, and (not saying youâre doing this) I think the absolutely disaster zone that was the sB1047 debate and discourse canât be fully attributed to e/âacc or a16z or something. I think the backlash Iâve seen to the AI Safety/âx-risk/âEA memeplex over the last few years should prompt anyone in these communities, especially those trying to influence policy of the worldâs most powerful state, to really consider Cromwellâs rule.
On this âyou will just in fact have pro-OS people mad at you, no matter how nicely your white papers are written.â I think thereâs some sense in which itâs true, but I think that thereâs a lot of contigency about just how mad people get, how mad they get, and whether other allies could have been made on the way. I think one of the reasons they got so bad is because previous work on AI Safety has understimated the socio-political sides of Alignment and Regulation.[1]
- ^
Again, not saying that this is referring to you in particular
I responded well to Richardâs call for More Co-operative AI Safety Strategies, and I like the call toward more sociopolitical thinking, since the Alignment problem really is a sociological one at heart (always has been). Things which help the community think along these lines are good imo, and I hope to share some of my own writing on this topic in the future.
Whether or not I agree with Richardâs personal politics or not is kinda beside the point to this as a message. Richardâs allowed to have his own views on things and other people are allowed to criticse this (I think David Mathersâ comment is directionally where I lean too). I will say that not appreciating arguments from open-source advocates, who are very concerned about the concentration of power from powerful AI, has lead to a completely unnecessary polarisation against the AI Safety community from it. I think, while some tensions do exist, it wasnât inevitable that itâd get as bad as it is now, and in the end it was a particularly self-defeating one. Again, by doing the kind of thinking Richard is advocating for (you donât have to co-sign with his solutions, heâs even calling for criticism in the post!), we can hopefully avoid these failures in the future.
On the bounties, the one that really interests me is the OpenAI board one. I feel like Iâve been living in a bizarro-world with EAs/âAI Safety People ever since it happened because it seemed such a collosal failure, either of legitimacy or strategy (most likely both), and itâs a key example of the âun-cooperative strategyâ that Richard is concerned about imo. The combination of extreme action and ~0 justification either externally or internally remains completely bemusing to me and was big wake-up call for my own perception of âAI Safetyâ as a brand. I donât think people can underestimate the second-impact effect this bad on both âAI Safetyâ and EA, coming about a year after FTX.
Piggybacking on this comment because I feel like the points have been well-covered already:
Given that the podcast is going to have a tigher focus on AGI, I wonder if the team is giving any considering to featuring more guests who present well-reasoned skepticism toward 80kâs current perspective (broadly understood). While some skeptics might be so sceptical of AGI or hostile to EA they wouldnât make good guests, I think there are many thoughtful experts who could present a counter-case that would make for a useful episode(s).
To me, this comes from a case for epistemic hygiene, especially given the prominence that the 80k podcast has. To outside observers, 80kâs recent pivot might appear less as âevidence-based updatingâ and more as âsurprising and suspicious convergenceâ without credible demonstrations that the team actually understands opposing perspectives and can respond to the obvious criticisms. I donât remember the podcast featuring many guests who present a counter-case to 80ks AGI-bullishness as opposed to marginal critiques, and I donât particularly remember those arguments/âperspectives being given much time or care.
Even if the 80k team is convinced by the evidence, I believe that many in both the EA community and 80kâs broader audience are not. From a strategic persuasion standpoint, even if you believe the evidence for transformative AI and x-risk is overwhelming, interviewing primarily those already also convinced within the AI Safety community will likely fail to persuade those who donât already find that community credible. Finally, thereâs also significant value in âpressure testingâ your position through engagement with thoughtful critics, especially if your theory of change involves persuading people who are either sceptical themselves or just unconvinced.
Some potential guests who could provide this perspective (note, I donât these 100% endorse the people below, but just that they point the direction of guests that might do a good job at the above):
Melanie Mitchell
François Chollet
Kenneth Stanley
Tan Zhi-Xuan
Nora Belrose
Nathan Lambert
Sarah Hooker
Timothy B. Lee
Krishnan Rohit
I donât really get the framing of this question.
I suspect, for any increment of time one could take through EAs existence, then there would have been more âharmâ done in the total rest of world during that time. EA simply isnât big enough to counteract the moral actions of the rest of the world. Wild animals suffer horribly, people die of preventable diseases etc constantly, formal wars and violent struggles occur affecting the lives of millions. There sheer scale of the world outweighs EA many, many times over.
So I suspect youâre making a more direct comparison to Musk/âDOGE/âPEPFAR? But again, I feel like anyone wielding using the awesome executive power of the United States Government should expect to have larger impacts on the world than EA.
I think this is downstream of a lot of confusion about what âEffective Altruismâ really means, and I realise I donât have a good definition any more. In fact, because all of the below can be criticised, it sort of explains why EA gets seemingly infinite criticism from all directions.
Is it explicit self-identification?
Is it explicit membership in a community?
Is it implicit membership in a community?
Is it if you get funded by OpenPhilanthropy?
Is it if you are interested or working in some particular field that is deemed âeffectiveâ?
Is it if you believe in totalising utilitarianism with no limits?
To always justify your actions with quantitative cost-effectiveness analyses where youâre chosen course of actions is the top ranked one?
Is it if you behave a certain way?
Because in many ways I donât count as EA based off the above. I certainly feel less like one than I have in a long time.
For example:
I think a lot of EAs assume that OP shares a lot of the same beliefs they do.
I donât know if this refers to some gestalt âbeliefâ than OP might have, or Dustinâs beliefs, or some kind of âintentional stanceâ regarding OPâs actions. While many EAs shared some beliefs (I guess) thereâs also a whole range of variance within EA itself, and the fundamental issue is that I donât know if thereâs something which can bind it all together.
I guess I think the question should be less âpublic clarification on the relationship between effective altruism and Open Philanthropyâ and more âwhat does âEffective Altruismâ mean in 2025?â
I mean I just donât take Ben to be a reasonable actor regarding his opinions on EA? I doubt youâll see him open up and fully explain a) who the people heâs arguing with are or b) what the explicit change in EA to an âNGO patronage networkâ was with names, details, public evidence of the above, and being willing to change his mind to counter-evidence.
He seems to have been related to Leverage Research, maybe in the original days?[1] And there was a big falling out there, any many people linked to original Leverage hate âEAâ with the fire of a thousand burning suns. Then he linked up with Samo Burja at Bismarck Analysis and also with Palladium, which definitely links him the emerging Thielian tech-right, kinda what I talk about here. (Ozzie also had a good LW comment about this here).
In the original tweet Emmett Shear replies, and then itâs spiralled into loads of fractal discussions, and Iâm still not really clear what Ben means. Maybe you can get more clarification in Twitter DMs rather than having an argument where heâll want to dig into his position publicly?
- ^
For the record, a double Leverage & Vassar connection seems pretty disqualifying to meâespecially as iâm very Bay sceptical anyway
- ^
I think the theory of change here is that the Abundance Agenda taking off in the US would provide an ideological frame for the Democratic Party to both a) get competitive in the races in needs to win power in the Executive & Legislature and b) have a framing that allows it to pursue good policies when in power, which then unlocks a lot of positive value elsewhere
It also answers the âwhy just the US?â question, though that seemed kind of obvious to me
And as for no cost-effectiveness calculation, it seems that this is the kind of systemic change many people in EA want to see![1] And itâs very hard to get accurate cost-effectiveness-analyses from those. But again, I donât know if thatâs also being too harsh to OP, as many longtermist organisations donât seem to publicly publish their CEAs apart from general reasoning like about âthe future could be very large and very goodâ
- ^
Maybe itâs not the exact flavour/âideology they want to see, but it does seem âsystemicâ to me
- ^
On Stepping away from the Forum and âEAâ
Iâm going to stop posting on the Forum for the foreseeable future[1]. Iâve learned a lot from reading the Forum as well as participating in it. I hope that other users have learned something from my contributions, even if itâs just a sharper understanding of where theyâre right and Iâm wrong! Iâm particularly proud of Whatâs in a GWWC Pin? and 5 Historical Case Studies for an EA in Decline.
Iâm not deleting the account so if you want to get in touch the best way is probably DM here with an alternative way to stay in contact. Iâm happy to discuss my reasons for leaving in more detail,[2] or find ways of collaborating on future projects.[3]
I donât have the data to hand but Iâd guess Iâm probably one of the higher percentile Forum users in recent years, so why such a seemingly sudden change? The reason is that itâs less a relation to the Forum and more due to my decision to orient away from âEAâ in my lifeâand being consistent there means stepping away from the Forum. I could write a whole post on the reasons I have for not feeling âEAâ, but some paragraph summaries are:
I donât really know what âEAâ means anymore: Thereâs always been a tension between the âphilosophy or movementâ framings of EA, so in practice itâs been used as a fuzzy label for sets of ideas or people. In practice, âEAâ seems to have been defined by its enemies for the past ~2.5 years, though I know CEA seems to change this. But I think this lack of clarity is actually a sign that EA doesnât really have a coherent identity at the moment. It doesnât seem right to strongly associate with something I canât clearly define.
I have increasing differences with âphilosophicalâ EA (as I understand it): This difference has been growing recently, and itâs a fairly long list of which Iâll only include a few things. I think viewing morality as about âthe bestâ instead of âdoing rightâ is a mistake, especially if the former leads to viewing morality as about global/âperspectiveless maximisation.[4] Iâm a virtue ethicist and not a consequentialist/âutilitarian. I think my special relationships to others in my life create important and partial moral obligations/âduties. I donât think Expected Value is the only or best way to make decisions for individuals or institutions. I think cluelessness/âKnightian uncertainty arguments defeat most of the cases for longtermism in practice. The various differences I have seem significant enough that I canât really claim to be EA philosophically unless âEAâ is drawn arbitrarily and trivially wide.
I also donât feel connected to the âmovementâ side of EA either: Iâm not personally or socially connected to much of EA. My friends and close personal relationships are not related to EA. While Iâm more professionally involved with AI Safety than before, I also make clear that my positions are fairly unorthodox in this space too. While Iâve done my own bit of defending EA on and off the Forum, I no longer feel the identification or need to. So Iâm starting to leave the various online EA spaces I am a part of one-by-one.[5] Given my limited connection to EA personally or philosophically, it seems odd to be part of the movement in ways that imply Iâm giving it more support than I actually do.
I think there are better ways for me to spend my time than engaging with EA: In the last ~6 months EA engagement hasnât made me happy. This has often been from seeing EA criticisms left unresponded to, and I donât want to be associated with something society views negatively if I donât support that thing! I also think that there are more interesting and fulfilling pathways for my life to pursue which are either orthogonal to EA, or outside the âorthodoxyâ of cause areas/âinterventions. Finally, I just think I spend too much time reading the Forum or EA Twitter, and going cold turkey would be a good way to reallocate my attention better. I donât think that the ârightâ thing for me to do in terms of doing the right thing, or for personal flourishing, is to engage with EA.
Overall Takeaway: I never really claimed the label âEAâ for myselfâit was never the basis of my identity and Iâve never really had the âtaking ideas seriouslyâ genie take over my life. But I think, given my differences, I want to put clearer distance between me and EA into the future.
Anyway, that turned out to be a fair bit longer than I intended! If you made to the end, then I wish you all the best in your future endeavours đ[6]
Definitely âfor nowâ, but possibly for longer. I havenât quite decided yet.
Though see below first
See the end of this comment for some ideas
Indeed, from a certain point of view EA could be seen as a philosophical version of Bostromâs paperclip monster. If there is a set definition of the good, and you want to maximise, then the right thing to do is paperclip the universe with your definition. My core commitment to pluralism views this as wrong, and makes me deeply suspicious of any philosophy which allows this, or directionally points towards it
I do, however, intend to continue with the GWWC Pledge for the foreseeable future
I also wish you the best if you didnât