Background in cognitive science. I run a workshop with the aim to teach methods to manage strong disagreements (including non-EA people). Also community building.
Interested in cyborgism and AIS via debate.
https://typhoon-salesman-018.notion.site/Date-me-doc-be69be79fb2c42ed8cd4d939b78a6869?pvs=4
I’ll give it a try !
Open question: would it be useful to frame this an “impact insurance” for people in impactful careers?
As in:
-I expect this goal to be good for the world (say ~100 WELLBYs in expectation)
-If I don’t achieve this goal, then I definitely owe something that’s a least comparably good for the world (say ~$200 to PureEarth (since I’ve chosen WELLBYs))
(or maybe at least half as good)
I think using it this way could help people who have a real hesitation between impactful work and impactful donations.
“Goes well for humans” (i.e for a very long time) worlds are mostly worlds where AGI is fully theoretically and empirically aligned with a CEV-shaped alignment target, which for me logically requires animal welfare. (I also currently believe those worlds to be implausible because no company seems focused on this)
I struggle to imagine any deliberative or reflective-preference oriented process that does not give the right answer to the animal welfare question. If it doesn’t care about non-human animals, then it means animals are not sentient, or that the CEV is misaligned with human interests and some humans will die because they don’t check the right boxes (and sentience isn’t a box), or that morality is weird and it’s actually fine to torture sentient beings (possible but implausible).
There are other worlds where “goes well for humans” means corrigible and aligned on some unaltered human values. In those worlds, I expect animals to take a blow on the short term, and possibly on the very long if the principal does not care about animal suffering. I also expect humanity to do other morally wrong things that they don’t suspect to be wrong, and die counterfactually way sooner.
Hopefully americans will take this as a strong signal that their administration is in complete support of mass surveillance and autonomous weapons. I could scarcely think of a clearer signal.
Edit : After reading different sides of the story, I’m actually less certain about that. But I’m still updating towards more probability that “unfettered access” meant “occasionally unlawful use” rather than “lawful use but there’s drama between us” given this reaction. I would certainly not have lended more probability if the administration hadn’t reacted so abruptly.
New Edit: after reading more, I’m pretty convinced that both issues (or any action that I’d personally consider falling into both categories) are pretty much legal in the US.
Exciting! I hope we’ll hear the same from AIM—Incubated Founding to Give initiatives in the coming years!
Strong upvoted.
I find this refreshing and going back to what EA is about, and would definitely point at it as “the sort of experience I expect an EA org to allow for”. I also appreciate the moral character that you’re showing up. I want more testimonies like this on the forum to help people get [back] in touch with this spirit.
I sometimes feel like I had partially lost my initial ambition, and your post corrected for some value drift. Thank you.
Note: You’re focusing on the English speaking sphere, yet e.g. the French EA community owes its existence in no insignificant part to two non-EA YouTubers. A third (very popular) one is soon adding himself to the list.
Sorry, I understand this is a bit confusing.
I was hesitant to spell it out, because I’m afraid of building a strawman:
My interpretation is that some people have an issue with non-self-oriented wishes or desires, because they can feel like virtue-signalling or guilt-tripping. Expressing things such as “I really want a world without malaria” can be interpreted as condoning the use of suffering as a negotiation tool.
I.e :
Step 1: People are suffering from malaria
Step 2: This prompts me to fight malaria
Step 3: Someone concludes that suffering causes me to help them
Step 4: They self-inflict suffering to them
Step 5: This prompts me to help them regardless
Step 6: The world is now made up of people who self-inflict suffering as a way to manipulate others, which suck.
I’m not sure this is an accurate reconstruction, but this is what I can do to the best of my abilities.
I’d rather not encourage arguing with this version of the argument, since I’m not a genuine proponent.
Be helpful, considerate, generous, genuine in your belief that (for example) malaria is bad and a world without malaria is a world you want to see.
Admittedly bitter take: you’d be surprised to learn this is far from consensual in some EA circles. I got surprising reactions for applying this example.
There seems to have been a surge for interest in AI Risk and Safety culminating on August 14th, far surpassing all other levels of interest in time.
I’m not sure what caused this. [Update : the EU Code of Practice ?] Google Trends lists seemingly topic-specific points (the “chain-of-thought as a fragile opportunity” paper from different AI Companies, an apparent interest in China about AI Safety) while one could intuitively bring up events that happened over the summer (several papers, the suicide case, etc).
Marcus, Austin, thank you so much! This is exactly the sort of tools Effective Giving Initiatives sorely lack whenever they’re asked about AI Safety (so far the answer was “well we spoke to an evaluator and they supported that org”). @Romain Barbe🔸 hopefully that’ll inspire you!
On my side, I’d be happy to compare that to the cost- effectiveness of reaching out to established YouTubers and encouraging them to talk about a specific topic. I guess it can turn out more cost-effective, per intervention, than a full-blown channel. I’m unwilling to discuss it at length but France has some pretty impressive examples.
Point of confusion:
I sometimes see linkpost or reposts on the forum, where I think to myself:
1-...is the person who’s posting standing by and defending what is said? Why are they posting this? Why don’t I have more context or commentary? If you do these sort of posts, what’s your rationale?
2-I sometimes disagree or think the epistemics are bad or the content of the post is clearly corroding the forum’s norms (often but not always hooligan-ish). I want to downvote them as hard as if they were genuinely written and endorsed as an EA forum post with an EA public in mind, despite the fact they’re presumably subject to softer moderation rules, and despite the fact that even my best criticism won’t reach the author if it’s a repost. Am I alone in thinking I’ll downvote anyhow and would encourage others to do so, in this context? If you disagree, what’s your take?
(e.g. DOGE cuts were a total surprise to most but are probably the single biggest event in global health this decade)
I share your take, and would add that this example is even more central than you seem to suggest. Not foreseeing this is a clear, traceable mistake. These decisions caused such a death toll that even moderate (bipartisan) efforts on that front would have been justified in terms of EV.
As a forum user, I’d say I think that it’s possible to discuss politics coldly (as long as moderation is stringent on tribal dynamics) and would appreciate this done more often (by the right people).
Hot take : France, Italy and Portugal should do the same thing.
There seems to be some cultural reluctance to EA, or analytic thinking, or weirdness, or ideas and arguments from the anglosphere more broadly in those countries. Under this view, simple translations aren’t enough, and it seems to be your perspective.
Some people argue this is not the case, however, and defend it’s rather structural. EA found the wrong audience in those countries, but even classical utilitarian EA would find supporters if oriented towards the right audience. They’d rather fund a university group in engineering and programming schools (as opposed to generalist universities) than write a whole new book.
What’s your take?
I’d be very grateful to have :
1-A precise example of systemic change given
2-Explanations for why e.g. ARMoR, Concentric Policies or Kooperation Global (or any other policy-focused org or conjunction thereof) don’t count as systemic change approaches (despite being definitely EA aligned)
3-An example of a tried and tested “solve the root cause” intervention, something I can look at and think “Oh, I want that but for GHD!”.
Another question : How did we come to the conclusion that a root cause existed for all/a large chunk of GHD issues ? This sounds like an extremely complex hypothesis to me. What evidence have we observed that is more probable to observe in case of a root cause rather than not ?
This helped clarify things a bit, thanks -not the least because I’ve interacted with some of you lately.
I’m still mostly skeptical because of the amount of implicit conjunctions (a solution that solves A & B & C & D & E seems less plausible to exist than several specialised solutions), how common it is for extremely effective ideas to be specialised (rather than the knock-on result of a general idea) and the vague similarity with “Great Idea” death spiral traits. All of this said, I’m in favor of keeping the discussion open. Fox mindset rules.
For those who need clarification, I think I understand four non-exclusive example avenues of what “solving the metacrisis” looks like (all the names are 100% made-up but useful for me to think about it):
1-”Systemism” postulates the solution has to do with complex system stuff. You really want to solve AI X-risk ? Don’t try to build OpenAI, says the systemist, instead, do :
1.1-One gigabrained single intervention that truly addresses all the problems at the same time (maybe « Citizen Assemblies but better »)
1.2-A conjunction of interventions that mutually reinforces each other in a feedback loop (maybe « Citizen Assemblies » + « UBI » + « Empowerment in social help »)
Will this intervention solve one or all problems ? Again, opinions diverge :
1.3-Each intervention should solve one problem -a bigger one, and to a better extent, than conventional solutions.
1.4-This intervention should solve all the problems
1.5-The intervention should solve one problem, but you can « copy-paste » it, it has high transferrability.
2-”Introspectivism” postulates the solution has to do with changing the way we relate to ourselves and the rest of the world. You really want to solve AI X-risk ? Again, don’t build OpenAI, but go meditate, learn NVC, use Holacracy.
3-”Integralism” postulates the solution is to incorporate criticism of all different paradigms. You really want to solve AI X-risk ? Make your plans consistent with Marxism, Heidegerian Phenomenology and Buddhism, and you’ll get there.
4-”Culturalism” postulates that cultural artifacts (workshops, books, a social movement, and/or memes in general) will suceed to change how people act and coordinate, such that reducing X-risks becomes feasible. Don’t try to build OpenAI, think about cultural artifacts -but not books about AI Risk, more like books about coordination and communication.
Separately, I think discussing disagreements around the meta-crisis is going to be hard.
Why? I think that there is a disparity in relevance heuristics among EA and… well, the rest of the world. EA has analytical and pragmatic relevance heuristics. “Meta-crisis people” have a social relevance heuristic, and some other streams of thoughts have a phenomenological relevance heuristic.
Think Cluster Headache. I think many people attempted to say that Cluster Headaches could matter a big deal. But they said stuff we (or at least, I) didn’t necessarily understand, like “it’s very intense. You become pain. You can’t understand”. Then decades later, someone says “maybe suffering is exponential, not linear, and CH is an intensity where this insight is very clear”. And then (along with other numerate considerations) we progressively started caring (or at least, I started caring).
All these systems can communicate with EA if and only if they succeed to formalize / pragmaticize themselves to some degree, and I personally think this is what people like Jan Kulveit, Andres G. Emilsson, or Richard Ngo are (inadvertently ?) doing. I’d suggest doing this for meta-crisis (math > dialogues) otherwise people may backfire.
I remember reading in comments from the world malaria day post that some malaria vaccines could turn out more cost-effective than bednets.
If that’s plausibly the case for Sanaria, I can’t but stand behind this ask (I’m sadly not in a position to help much more than that).
I personally feel the website is more engaging, but unexpectedly, I also feel my own conception of EA is more accurately represented !
I’d be excited to experiment with some of these insights, and these methods in general, for other community building practices.
As an ex-group organiser, I feel that my fallback plans have just been described with extreme precision.
Thank you for sharing my old-age frustration !
More seriously, I agree that this is a recurring problem, and I’m not sure where people want to put the gradient between not fearing to oppose consensus and lacking pedagogy.
A thing I recommend now is to ask your best accessible version of Claude to train you on Street Epistemology, Smart Politics and Deep Canvassing for those specific scenarios in a 3 weeks program based on deliberate practice. It already gives excellent foundations for these situations.