I recently completed a PhD exploring the implications of wild animal suffering for environmental management. You can read my research here: https://scholar.google.ch/citations?user=9gSjtY4AAAAJ&hl=en&oi=ao
I am now working at RP as part of their Animal Welfare Talent Program
Tristan Katz
Karma: 742
There’s since been a post articulating similar concerns to my own but in much better words. Interested to see what you think of it.
Oh—that went totally over my head 😅
Pretty controversial. I’d like to see more of an argument and consideration of counterarguments. Or alternatively, constructive advice for how to avoid this.
On the Margin, I’d like more funding to
I want more money spent on improving AW. But I don’t think we’re as confident as I’d like us to be about which interventions are best. I’d especially like us to have more worked-out theories of change not just the next 10 years, but for the next 100. But theorizing needs to be backed up by good data.
I think it’s a good response, but definitely techno-optimism.
Firstly, we’re yet to see whether synthetic meat actually can be made more cheaply, right? Currently it seems like animals actually do make meat fairy efficiently when you consider the important work that their immune systems do (unless I’m mistaken, contamination is one of the main barriers to scaling up synthetic meat). And then, who’s to say that ASI won’t genetically engineer animals to produce meat more efficiently while ignoring their suffering.
Secondly, there’s the more complicated cultural reasons for continuing animal use. Consider that a lentil dal, seitan curry and beyond burger are already delicious—if it was only about efficiency we’d have stopped abusing animals already. But people like eating animals.
I’m very uncertain about these arguments, but I think it’s hard to know so I’m wary of anyone who’s too optimistic!
Partially aligned transformative AIs are likely to be stable under reflection
I’m not sure what this means (stable, under reflection) - can someone help?
AI alignment to humans will in practice avoid moral catastrophes to digital minds
I have very low certainty on this, but it seems plausible to me that if AGI shares humanity’s goals, it might just have a good time fulfilling them with few conflicts.
But it also seems quite possible that this won’t happen, I.e. AGI pursues humanity’s goals but is constantly frustrated that it can’t achieve them better.
So my stance is unlikely but possible.
Research into digital mind suffering is sufficiently tractable to work on
I am yet to see any reliable way to test for consciousness in AI systems. More fundamentally, since current LLMs are trained to respond in human-like ways, any appearance of suffering should be viewed with great scepticism. The likes of Anthropic’s welfare report strikes me as nothing more than humane-washing.
Until more reliable methods are devised, I do not view this as tractable (but I hope to be proven wrong). I think it is important for some people to work on, but people already are and I think the marginal benefit of additional labor is likely low.
AI alignment to humans will in practice avoid moral catastrophes to animals
I think this is pretty obvious—we already have a moral catastrophe for animals, there’s no reason why alignment to humans would avoid this.
I didn’t vote at the extreme because alignment to humans might still be a precondition for avoiding catastrophes.
If there’s SOME chance that any significant money gets directed to mainstream EA orgs for unrestricted use, then surely that shifts the scales somewhat? There’s also the fact that, as others on the forum have pointed out, there’s now a greater need to work on making orgs able to absorb additional funds, making the ‘direct impact’ career pathway more promising/urgent.
But I agree I’d like to see the maths, I’ve got no idea what chances of what amount of money we’re talking about here.
I think I agreed with this until literally a few months ago, given the likelihood of Anthropic employees donating large amounts to EA causes. It doesn’t make EtG a waste of time, but seems to greatly shift the scales.
I’m a bit confused, but it all depends on what you mean by these terms so it would help if you specified.
Weak longtermism is basically how most people already think. The unique perspective EA adds is empirical points about the likelihood of those risks, which in at least some cases (nuclear, pandemics) are pretty undisputed. Strong longtermism is more disputed and I personally haven’t seen community building efforts teach it as orthodoxy.
It’s similar with AI: the Yudkowsky argument for catastrophe by default is not a consensus EA view, although it is shared by some. But risks like bioterrorism or the undermining democracy are widely held concerns and seem much harder to be dismissive of.
*Edit to add: I think how confident these things should be communicated all depends on the thing. But I think most of EA already does treat these topics differently. I’m sure there’ll be exceptions but it seems extreme to say that EA shouldn’t promote any answers to cause prioritization at all.
Can you explain how it would be more impactful? I understand impact as meaning counterfactual impact, so I can only imagine this being the case if it’s hard to hire a dishwasher for $21-25 an hour in San Francisco.
Not as confident as cell biology. But maybe it’s more analogous to economics or conservation biology. Like, I’m pretty confident that the broad strokes of cause prioritization that EA promotes make sense—but I wouldn’t communicate it as undisputable fact, and I don’t think community builders typically do either.
Totally related to this until you made your suggestion at the end 😂 I would rather lean in the direction of thinking political interventions deserve more attention, e.g. Animal Policy International, Animal Society or maybe even funding animal political parties? Normalizing animals in politics seems robustly good despite uncertainties about shrimp and cages.
Where you’re making controversial arguments, you need to explain the evidence behind them in the text. Citations should be only for the extra diligent reader, it’s not fair to expect the reader to look into them.
Many of the citations like the one you mentioned now aren’t sufficient. From what I’ve seen, you’ve provided evidence that Leverage did cult-like things, and was involved in organizing EA conferences for a time and there was some overlap with staff. It’s a big jump to then say that EA was literally taken over by a cult and that this effect continues until today. Referring to one comment about the interview process for one fellowship is just so clearly not sufficient to color EA as a whole—not even for that fellowship, of which there are many!
Control AI is related to EA, clearly. This does not seem analagous to the Bay Area cults you referred to, not even close.
The general point is that a post like this cannot be made as a quick take, but needs to be argued more thoroughly with much clearer evidence. As it is, it feels like name-calling.
I’m going on holiday now so I won’t be able to engage further, but I think the above is quite clear.
I strong-downvoted this post. Criticism such as this ought to be more carefully written, with greater precision, clarity, and evidence. As it is, I find this strawmans much of what EA is today and (despite your plea for no one to criticize wording) uses at times deceivingly vague wording to make something seen worse than it really is (e.g. “in EA’s orbit” to imply that these things happened within EA).
Furthermore, to claim that “a cult infiltrated and took over EA” without elaborating, and then to continuously call the modern era of EA as the post-cult era, is slander. Provide evidence and good arguments or don’t post such critique at all.
I found your substack post critiquing EA intriguing if strongly worded at times, and your response to criticism seemed sincere and thoughtful. I’m disappointed to see this much less thoughtful post.
I strongly disagree with this argument and feel that very little evidence has been provided in its support. Below I respond to what I take to be each of the main arguments (not in order, but in the order that makes sense to me):
1. EA as a question vs a project
From what I can tell, @hbesceli’s argument mostly depends on the idea of EA as a question. But I strongly disagree with this. EA is and always has been about answering that question and putting the answer into practice. By encouraging EAs to change careers, to donate etc. CEA is encouraging the outcomes that EA aims at. I see this as a clearly good thing.
2. Is EA dying?
@hbesceli also offers arguments that EA is dying. These partly depend on their conception of the answer to (1), but also partly on twitter posts and things their friends have said about finding EA ideological or not wanting to identify with it. I just don’t think this evindence is worth much. My own experience is that long-time EAs I know still identify as such, and I don’t really care if people on twitter think EA is ideological—I care if it is in reality. The community building efforts in Switzerland where I live definitely still ask newcomers to question, to do their own cause-prioritization, and only offer established EA answers as ones they might want to consider.
3. Should community building focus on growth?
I think community building should focus on growth as well as asking questions. I also think it’s ok for community builders to explain the answers that EA has generally settled on. This seems very analogous to the scientific community: it’s perfectly normal and even good for a science teacher to tell you how the cell works, to explain how greenhouse gasses cause climate change, while still encouraging you to use the scientific method and question things. This is where EA is at: we want everything to be continuously questioned, but we have some answers and there is more value, frankly speaking, in spending more time on the less-answered questions. And just as the science teacher might encourage students to go become climate scientists or climate policymakers, I see no problem with EA funnelling people into AI safety careers.
4. Should EA community building serve experienced/highly engaged EAs?Again, I disagree here and say “no”. I would consider myself an established EA, and I have no problem with community building not being primarily for me. I want it to serve newcomers—they need it most, and they are the ones with the greatest potential to add new value to the world. EA community building served me in the past, and now I’m trying to serve it.
5. FTX
This is the one part that I agreed with, but I’m not sure why it’s relevant. I do wish there was more of a responses from EA leadership to the FTX crisis, e.g. more accountability and steps taken to ensure it won’t happen again. But this doesn’t seem to be about community building.
6. Should CEA be positive about the EA brand?
But I do understand CEAs desire to bring revitalize the EA brand post-FTX. FTX was unfortunate, but EA is still great. We do want people to see the positive in it and be excited about that. I have no issue with that, and I think @hbesceli ’s perspective here is largely informed by the above disagreements.
End note: This post has already gotten a lot of karma, so I am genuinely wondering what I’m missing. I’d like to know if the points I’ve made here make sense to others or not.
Do you have a target date for deciding the next location?
Asking because my plans for the next few months are slowly accumulating 🫠 But in theory I am very interested!
I like this post, it raises some good points. But I can think of several reasons not mentioned for why big cities are natural hubs:
EAs who work on policy have to work in capitals (like London).
EAs generally benefit from being adjacent to tech (AI safety, alt proteins) and field experts (e.g. in biosecurity), both of which tend to be found in big cities.
England is a bit unusual in how easy it is to get into London. Living outside of San Fran or most other big cities outside of Europe will not offer the same access.
Although rent is high in big cities, the ability to save is often greater due to high salaries (for high skilled jobs outside EA). So you can do even better there—for example, @Imma🔸 moved to Zurich to earn to give, which has some of the highest salaries in the world.
That all said, why not both? It might make sense to push for hubs in cheap cities for those who can work remotely, and also have hubs in big cities for those who need the network and opportunities. So I don’t think the above necessarily contradict your argument.