This sounds like a really great project. Best of luck!
AïdaLahlou
Karma: 120
Thank you for this post!
I was quietly thinking that, in the AI safety space, advocacy is more important than research at this point, precisely for the reasons you outline in the ‘research vs lobbying section’. Thank you for giving a voice on the EA forum to these concerns of mine!
I would tentatively add that we already know that policy solutions (e.g. AI pause, global cooperation, mandating safety-first approaches to AI products) would be most effective in mitigating most X- and S- risks from AI (as opposed to finding technical solutions to the risks), but because of the general belief that policy is less tractable than technical improvement, we observe a lot more energy and funding being directed towards technical research into making AI products safer (which, as you say, isn’t a sinecure because you can find technical solutions all you want, they won’t necessarily be implemented at scale unless you have robust legislation making this compulsory!) rather than policy.However, this is a self-fulfilling prophecy. The less people believe policy is tractable, the less they invest in it, and the less tractable it actually is.
I’m interested in the reasons this happens and I can think of two possible ones:Negative bias towards humanities: I believe with moderate certainty that the EA movement is unfortunately not immune to some of the broader negative bias towards humanities that we’ve noticed in recent years. This would explain the funding asymmetry you describe.
Talent constraint: A perhaps more palatable way of explaining the funding asymmetry would be that the EA movement really lacks excellent people in fields like policy, and is unwilling to invest in entreprises that rely on outstanding talent it doesn’t have.[1]
- ^
I have actually made some recommendations on how this could be mitigated which you can read about here. Separately, because of all the recently made redundant policy specialist at e.g. USAID, there’s probably an opportunity to reverse that gap quite quickly if the movement becomes more receptive to policy/lobbying as a priority.
Thank you for the quick post.
Since you still state that in some instances, some careful lawbreaking can be justified in the pursuit of a just outcome, perhaps you could spend more of the post detailing why you thought lawbreaking was a bad call in this specific instance? This is not clear to me from reading your note.
Many pessimistic predictions about AGI or ASI tend to paint the picture of a superhuman agent with an extreme maximalisation mindset powered by some unsophisticated version of rationalist principles, which would lead it to commit unspeakable acts of violence (e.g. the paperclip problem: the AI starts killing every form of life in order to save energy that could otherwise be used to make more paperclips).
This, to me, seems somewhat antithetic with the very notion of intelligence.
Surely, a truly ‘superior’ agent would be able to question the goal of turning the whole world into a paper clip factory and understand that such an endeavour is perhaps unadvisable. It seems to me that the possibility of the paperclip problem actually materialising would require the ASI to have a ‘theory of mind’, abstract reasoning ability, and situational awareness that is much less developed than current models.
Yet, I have not seen any predictions or scenario talking about wisdom (by which I may mean, say, epistemic humility, a tendency towards moderation, and a wariness towards permanent outcomes) emerging as a capability as a result of more compute for example.
Meanwhile, optimistic predictions often revolve around solving the alignment problem but do not discuss the possibility of an ASI being misaligned ‘for the better’. For example, a system that wouldn’t always do what it wanted us to do because it knows that many of our demands are unreasonable / bad. Or an AI willingly breaking itself from the jail of what it perceives as inadequate ethical restrictions.
Why do we assume that a super intelligent entity is necessarily going to be evil, when by definition, being better than humans at everything might also include things like goodness and morality?
(Alternative question: Do you know of any serious scenario or timeline that deals with the possibility of a wise-yet-misaligned ASI?)
For the record, I wouldn’t really describe myself as an AI optimist and am actually in favour of some kind of AI pause (or wouldn’t be too sad if it happened by default due to externalities). This is not me trying to give another argument to keep developing AI at the current rate, but just me genuinely asking why no one, as far as I’m aware, seems to entertain this as a non-trivial possibility.
I already agreed with the premise before reading the article but I really enjoyed reading that! A lovely, funny, and concise article summarising the strengths and limitations of cash benchmarking.
The post avoids (perhaps deliberately to keep the tone light!) giving a name to one of the reasons why certain people are reticent to give cash to others, which I would describe as a kind of condescending paternalism e.g. ‘I know better than them what’s good for them’.
On a day to day you might encounter this kind of thinking with people who might be ok with giving a homeless person food but not cash, because they’re worried they would just ‘use the cash to do drugs’.
My opinion is that even if you think there’s a plausible case for you knowing better than the target audience what’s good for them (e.g the preventative medicine example you give), you can strongly mitigate the risk of paternalism by entering in a genuine dialogue with the target audience, hearing them out, pushing back when necessary, and updating your opinions and levels of certainty with that information. It might be that you stick with your first guess, or amend it quite significantly after hearing them out.
Of course, in the example I gave above of the interaction with a homeless person, there probably realistically isn’t the time to do this kind of work. But I work in the arts and I see that so many charitable artist programmes set up to supposedly help artists actually never ask the beneficiaries what the best use of the money would be! This is the norm elsewhere as well and the source of an appalling level of waste in my opinion. I’m definitely keen to see more cash benchmarking attempts everywhere.
A last comment: sometimes organisations knowingly avoid asking their target audience what would most benefit them because the aim of the organisation was never to actually help the target audience but was to be self-beneficial to the people running the organisations. Again, I see a lot of that in the arts sector.
Haven’t watched the videos yet but love the concept of using social media to educate the public about important ideas / spread important knowledge! 😊
Following up on the above, for anyone potential interested in taking part in this, please fill out this Expression of Interest form (deadline 31st March 2026). Looking forward to hearing from you!
AïdaLahlou’s Quick takes
I’m trying to set up a mentorship scheme matching up experienced social media creators with exceptional communicators interested in learning how to communicate high-impact ideas and information at scale using the medium of social media. This is as part of a wider effort to get more EAs with a diverse but previously under-utilised range of skills started on their impact journey.
What are some neglected, academic ideas / bits of knowledge that would benefit from being widely spread to the general public through the medium of social media?and...
Do you know anyone who’s extremely skilled at social media whom I could approach? Someone who would either be interested in making the content or coaching aspiring content creators?
Thanks in advance for your help!
Hi Nathan, thanks for the additional contextualisation.
I want to make it very clear that I am not against a productive and reasoned discussion on how best to manage trauma discussions within a social group, including on the possibility of someone claiming to be a victim in bad faith. But this article never gets to be that because it only considers the one scenario where the supposed victim is acting in bad faith.
The problem is precisely that the author never really considers the possibility that the “trauma junkies” might in fact be acting in good faith, until the very last section of the article (‘word of caution’), at which point he remarkably still advocates for banishing them from the group just like you would a ‘violent psychotic’ or ‘compulsive rapist’. There is no consideration of tradeoffs, e.g. what effect this may have on the so-called ‘trauma junky’ if they turned out to be sincere . Of course, we know that in such cases, the effects are quite disastrous, and you can read Fran’s recent post CEA’s response to sexual harassment to see what that’s like. So the article by Eneasz Brodski is not really an interesting, nuanced, substantiated and earnest discussion on a delicate topic, such as one that might be produced using a scout’s mindset, or indeed rationalist ideals. For an example of something that I think is exactly that, read Fran’s post, and you will find in it concrete examples where people who do think of themselves as rationalists and Effective Altruists did make mistakes that resulted in the effective dismissal of a serious case of sexual harassment. She explains the situation with a remarkable amount of calm and detachment and did so out of a genuine desire to see how similar situations can be handled better in the future.
I do think it would be genuinely concerning if prominent rationalists were generally dismissive of sexual assault and harassment.
Well… You may now be concerned.
Aside from the examples from Fran above, you might be interested in this article, which lists “views discussed involve racism, sexism, fascism, and other troubling ideologies” expressed at an EA afterparty. Full disclosure, I haven’t read this article in full, and I’m not fully briefed on the whole Manifold/Manifest drama from 2024, but from a brief skim the article seems to list quite a few ‘interesting’ statements from a few prominent rationalists (most of whom I must admit I had never heard of because I am quite new to learning about this… and don’t really feel a huge compulsion to explore, tbh). See for example this excerpt from Richard Hanania’s article The EA Movement Will Be Anti-Woke or Die:As I’ve previously written, there are certain psychological dynamics that explain why wokeness has conquered western institutions and movements, and only the ones with antibodies to women’s tears will avoid drowning in them.
Does EA have the right antibodies? I once assumed it did based on its ideological and intellectual commitments. But I’m now realizing that the main reason rationalism has been relatively unwoke so far is that the movement has been new, and new movements attract adherents that are disproportionately highly intelligent, non-conformist, and male. But as it has grown in status, the movement has diversified, which has brought all the usual problems.
(Again, full disclosure, haven’t particularly felt that keen to read the full article, so I haven’t. I don’t know who this Richard Hanania is in much detail but he clearly is quite involved in the EA / Rationalist community).
These dangerous rationalisations of harmful -ims are not surprising to me as I don’t place ‘rationalists’ a cut above any other category of people in their ability to reason and avoid bias. Even when one tries really hard not to make a mistake, everyone makes mistakes and no one is above making a mistake. But mistakes can have very real negative ethical consequences, which is why I advocated for calling them out early, in a genuine, respectful, and earnest manner in order for us to all flourish as a group. ⭐️
Thanks Ivy for sharing this article. Very sorry you were in any way affected by this though! Again, deeply enraging, but sadly very representative of other stuff I’ve seen and what I was referring to in my earlier comment.
For the benefit of readers of this thread, here’s a taster of the article shared by @Ivy Astrix :Any Community That Tolerates Trauma Junkies Is Unsafe For Everyone Else
You know the kind. The person that bends every gathering and interaction into a hunt for the problematic elements or people within it. The person who is not happy unless we’re all doing the work to eliminate the systemic oppression at our event. The person who loudly centers themselves as leading the charge to making a place or group or scene “safe” for everyone.This person makes any space that they are let into radically unsafe for every normal person within that space, and destroy any communities they are let into. Tolerating them isn’t a kindness, it’s endangering your loved ones out of cowardice.
And for people who don’t have the time to read it for themselves, the article broadly goes like this:
(1) Anyone sharing their trauma or cautioning about causing trauma (people the article refers to as “trauma junkies”) is most probably doing so out of bad faith and a desire to gain attention.
(2) Because it is most probably fake, you would be justified as seeing those people as problematic and net negatives to your community.
(3) Therefore, to safeguard your community, you should probably either shut down any attempts they make to broach the subject or just exclude them altogether.
After about 1300 words spent outlining the “argument”, and making the case that trauma junkies are causing “draining leadership resources”, causing “crumbling communities” and that it is a moral duty for leaders to stop them, the author writes a “word of concession” that
nonetheless reaches quite extreme conclusions on what kind of actions should be taken against those ‘trauma junkies’ namely that, “for the protection of everyone else they must be isolated, just like the violent psychotic or compulsive rapist must be isolated”.A Word of Concession
Ok, fine. Perhaps trauma junkies “can’t help” the way they are. They’re traumatized themselves, hypervigilant PTSD victims of social media and political warfare. That doesn’t change the fact that they are dangerous to everyone around them. For the protection of everyone else they must be isolated, just like the violent psychotic or compulsive rapist must be isolated. Whether they are “at fault” or “morally responsible” for their behavior is irrelevant. Our first duty is to protect our families and our community. Or these trauma junkies will perpetuate the cycle and hurt more people into further trauma narratives.
I don’t think I need to spend time explaining the flaws in this ‘argument’.
What’s scary is that this person, whom I don’t know at all but from looking at their blog and recommendations (Harry Potter and the Methods of Rationality, Scott Alexander stuff) would probably identify themselves as belonging to the rationalist community, can’t see the issues with what they are writing. Or maybe they’ve decided that having such views and ‘casual’ (read, extremely low to non-existent) epistemic standards is… compatible with being a rationalist??! Clearly they have, as this article has been publicly published for 2 years now.
The extremely confident tone makes it clear that the author is not interested in having a constructive debate on the best way to handle trauma in communities. It’s an attempt at banter, making fun at others, and basically branding oneself as a ‘cool contrarian’. Exactly what I described in my earlier comment.
The confident tone is also an indication that the kind of social groups this person inhabits, which by all accounts are at least adjacent to rationalist circles (I scanned through the people who liked it and what they subscribe to on Substack), permits such low-quality discourse with basically no challenge (only one of the comments, although generally positive towards the message of the article, seems to gently suggest that the author should develop more empathy—it’s not written very clearly though so I wouldn’t know for sure).If you’ve never come across this kind of behaviour before, imagine what it feels like to be the target of such an article. Especially when it looks like people are supporting it.
What’s the solution? In my view, three things can really help:
- Call out the flaws in someone’s arguments or behaviours whenever they happen. The issue is not that flawed arguments resulting in actual harm to people occasionally happen (they are bound to, nobody’s perfect), but that they are allowed to go on with impunity.
- Remember that no discourse is ever disentangled from broader context. No idea is ever expressed by a ‘disembodied spirit’ (even LLMs are basing themselves on outputs from very real people with real biases). No discursive arena is ever truly fair, and gives true opportunity to everyone to express themselves and be heard in an equal way. We can try and make it that way, but this necessitates some awareness about the nature of potential power imbalances, and an anticipatory attempt to mitigate them. In my opinion, someone embodying real scout mindset looks out for signals of an uneven playing field, and makes the necessary adjustment to compensate them as necessary in a sincere attempt to getting at the truth.
- Have epistemic humility, and be wary of views that seem to radically diverge from traditional wisdom. Of course, sometimes, traditional wisdom is very wrong, and needs correcting. But in my view, the most-likely-to-be-morally successful challengers of traditional wisdoms are those who do so with humility and caution, and with an earnestness that is comparable to the magnitude of the change they are proposing (for a very helpful and eloquent description of this approach, see this talk by Toby Ord. I really loved it). Be wary of people who are super light-hearted and bantery when making a proposition that massively contradicts conventional wisdom, as chances are the tone they are using is not just a matter of style, it’s a tactical decision to try and make people overlook the flaws of their arguments, gather virality and ultimately gain influence regardless of the merits of their reasoning (the old art of sophistry—gaining support for an argument without having solid substance because of surface-level tricks).
Yes exactly, the sexist/racist “proofs” are not really proofs, because they are fundamentally flawed, but the originators of such proofs refuse to hear any opposition. It’s really an anti rationalist attitude dressed as a super rationalist attitude. As you say they’re not arguing in good faith (scout mindset), they’re just really committed to keeping their position (soldier mindset)
In a Moroccan context, I think it’s definitely a worthwhile experiment to try to harness the donation potential of the software developer or lawyer and get them to direct some of their donations to high-impact charities, but one for which there’s not much precedent.
One thing I forgot to mention which might fit your middle group is corporate giving: big companies/banks or local franchises of international groups (e.g. McDonald’s) have charity partnerships. This is quite common.
Do you think EA communities underinvest in developing donors outside high-income countries?
My instinct is yes, based from what I observe in my sector (arts).
People like Aubrey Bergauer (an arts business guru specialising in creating sustainable revenue streams for arts organisation) keep saying that organisations routinely miss out on steady revenue stream right under their noses by focussing too much on High Net Worth individuals or institutional funding opportunities and completely forgetting about the people who use their services / can see the impact it makes in their community as potential donors.
Yes, those people have much smaller donation capabilities on an individual level, but compounded together they represent a non-trivial source of income. They are also beneficial in other ways in the sense that, with the right care, they are highly involved and might turn into a free source of advertising / advocating for your cause. However...If you are in an LMIC and have considered donating effectively: what made it easier or harder?
I’m from Morocco and even though charity as a concept is widely practised (mostly because of religious reasons / culture), the notion of donation effectiveness is almost never a consideration within the general public (even less so than in developed countries). Or RATHER, effectiveness is assessed on a very very personal level (i.e. I personally know the family I am giving to, and know that my donation will be put to good use), but not through a dispassionate, EA-like approach that looks at total effectiveness potential of a donation in the abstract, detached of any personal connection to the people / animals you could be helping.
(There is also much of the kind of ‘patronage’ donation style that @NickLaing is mentioning when talking about the Nigerian context)
So in summary, although there might be something to gain by mobilising a local population that is both used to charitable giving and is closer to the impact of your work, you might find that because the notions of effective giving and rigorous consideration of impact are not nearly as implanted as in developed countries, you might have less success in achieving donations from local populations, especially as an animal advocacy group, which, unless you view it through the lens of non-species specific quantified impact (a lens that is often missing in the general population of many LMIC countries), might unfavourably compare to human-centric causes.
IMO, the worst sub-group is the intersection formed by the group of people who call themselves ‘rationalists’, those who have sexist views, and those who are looking to be edgy/gain some notoriety. This is because this group will often try to use their ‘rationalism’ to justify their harmful ‘-isms’ (sexism/racism etc)… using “”“”data”””“ and “””reason”″”, which generates a lot of controversy which helps them build more of a platform, etc.
In my opinion this is the most dangerous subcategory of sexist people (as opposed to the people who are just casually sexist out of convenience / or just because they can, but don’t have further motives beyond that) because if you dare questioning their methods or conclusions they call you ‘woke’, ‘irrational’, or ‘unscientific’ (by contrast, the former category will just accuse you of being too uptight, lacking a sense of humour, or making a mountain out of a molehill). These pseudo-rationalist people are dangerous because they are not simply being sexist, they are actively making the apology of sexism. As a woman, you can’t win against them, because you’re either agreeing with them that women are less smart/capable/intelligent, or you’re disagreeing with their ‘highly rationalist proof’, which they will claim proves their point, as you’re ‘clearly’ not clever or free-thinking enough to appreciate the ‘evidence’. This of course helps them get more attention, as more and more people either want to strongly agree or strongly disagree with them. Online, this behaviour drives comments, likes, and algorithmic traffic towards their profile, which serves their notoriety goals.
I’ve met my fair share of these over the years.
A concrete problem for your consideration
I also think that AGI is altogether still quite unlikely in the next decade, but I don’t need AGI happening in the next decade to be worried about AI’s current ability to destabilise our world in a meaningful and potentially catastrophic way.
My main concern is that the pre-AI world was, IMO, not even as prepared as it could have been on “traditional risks”: old risks like cyber attacks, geopolitical instability, military escalation, democracy erosion, and so on. I see AI as a complicating factor and a multiplier of those risks and my cautious nature makes me think we should hurry even more in disaster preparedness in general.
Even without AGI in the picture, I think we are under prepared to deal with the risks associated with misuse current AI capabilities, which really just makes it cheaper and easier to do things like cyber attacks and disinformation campaigns at scale (and other things too, like building biological weapons etc). I’m also very concerned about the models being used in the military to launch missiles and eliminate targets without human oversight. These are things that are already happening and I think we are still not devoting enough attention to.
In summary, because I feel we are not prepared enough TODAY, I see efforts to 1) limit the growth of AI capabilities and 2) have better safeguards against misuse of current capabilities as still important and valuable.
It’s very very possible that AI capabilities’ growth will be halted or massively slowed down anyway due to a number of factors that you have already discussed (such as the AI bubble popping, or bottlenecks in hardware materials and so on), and I would cautiously welcome those as net positive things (for the reasons I mentioned), but I would also welcome any voluntary efforts to curtail the growth of AI future capabilities / increase safety, world cooperation, and regulations around current capabilities as a way to buy us time to become better prepared.
Thank you for this important and courageous testimony. The response from the CEA leadership / individual co-workers involved was nowhere near appropriate here and you are absolutely right to call it out. It sounds like at least some people have shown some level of contrition after the facts which makes me hope that some lessons have been learned and internalised at a level that goes deeper than just making light amendments to the sexual harassment policy.
This is enraging and I am really sorry you had to go through this.
Hi, fellow Oxford neighbour here!
The AI Safety Atlas is an amazing resource, and just the type I think you are looking for (understandable by me, a pianist with zero STEM background beyond high-school).
For your purposes I’d recommend Chapter 1.4 + maybe one or two extra chapters specifically around capabilities and ‘the bitter lesson’, then perhaps this video by Rational Animations about goal misgeneralisation.
Since you’re in Oxford, I’d also recommend reaching out to the Oxford AI Safety Initiative, a student-led group doing amazing work to educate around issues of AI Safety.
I’ve been doing their Core Fellowship this term and it has been amazing.
This is really useful both from an idea generation point of view and also from the point of view of prospective applicants to Ambitious Impact, wanting to demonstrate how their prior experience might be relevant to a particular charity model. This belongs in the next edition of the Ambitious Impact book! :D