I used to think something along those lines, but given what happened since the release of Opus 4.5 in November 2025, I now personally believe that EA as a whole is not focusing on AI enough.
I also believe that 80,000 Hours doesn't get enough credit for having pushed for focusing on what's most important since well before it was so clear and popular, and people like me who were complaining about them just telling everyone to work in AI should apologize and update sooner rather than later.
Some thoughts on your points:
1. There’s no longer a good resource to introduce people to EA.
I agree that 80k was possibly the best, but the two you link are also good, there's also https://forum.effectivealtruism.org/handbookwhich imho is still decent despite (or maybe thanks to?) being unmaintained
2. EA should be about diversity of thought
I personally strongly disagree with this, EA should be about helping others effectively. “the science of doing good” should reject "alternative science".
I'm not claiming that there is consensus among reasonable people that trying to influence the potential upcoming transition in a positive way is the best way to help others, but you shouldn't embrace diversity of views for its own sake.
3. Less people are discovering EA now.
This seems less important than people moving into careers that are better for the world, or donating more money, or influencing policy, or many other things. So I think it makes sense that more and more organizations are focusing on results rather than on EA's growth. I also don't think it's true (the US Department of War tweeted about effective altruism, The Economist had an article on EA on its front page, there's a major Netflix series coming out in November, ...)
Also, lots of people are still getting into non-AI EA from 80k. In the past 6 months 24 people signed the 10% pledge with 80k as a primary source. Given the massive interest in AI and 80k's AI focus, I expect that tons of people are discovering 80k and through it EA.
4. 80,000 Hours’ content is increasingly bizarre, which is bad for EA.
I think CEA is focusing on having very simplified, normie-friendly pages, without mentioning anything weird like the fact that most life in the next century might be of non-biological origin and deeply uncertain about its own moral status. I think it makes sense for CEA's content to focus on the EA brand and for 80k on how to do the most good with your career (see below on focus)
I would feel pretty uncomfortable endorsing these ideas at a party, and I imagine you would be too.
I think there's a very important difference between not endorsing those ideas because you disagree with them and not endorsing them because they're too weird or far from the social norm. I also feel uncomfortable mentioning significant donations and animal welfare, but I think that positively correlates with those things being very important to talk more about, because I deeply believe the related ideas to be true, and the discomfort is due to mistaken norms and beliefs.
It’s important to be willing to entertain ideas even if they sound a bit unusual, but putting a large spotlight on the movement’s most bizarre ideas is a bad idea. [...] it implicitly endorses the idea that the more strange an idea is, the more likely it is to be impactful. This turns EA as a project into a search for increasingly bizarre ideas by setting bizarreness as a goal rather than a happenstance.
Those are very far from the movement's most bizarre ideas, they are all reasonable things to focus on, and 80k's reasoning for focusing on those is well laid out, they're obviously not writing them just because they are bizarre.
To be clear I think it would be weird and bad if CEA redefined "effective altruism" to be about positively influencing this potential transition, instead of being about trying to help others from a broad range of perspectives, but I think what 80k is doing is very good for EA and for the world, and many more of us should be doing similar things.
I'm still quite uncertain about the topic so I don't feel committed to the view that AI takeover is likely.
I personally don't think you need to commit to the view that AI takeover is likely to encourage more focus on influencing upcoming transformative technology, you just need to believe that it's reasonably likely to be transformative and that there are reasonable enough theories of change to try to influence that transition in a more positive way.
Separately from considerations about the relative importance of AI, I also think from the outside it can be very hard to appreciate the value of focusing (see e.g. GWWC retiring 10 initiatives, which internally we called "Project Focus", led to clear significant growth in our key metrics and was extremely successful).
By focusing more on their respective priorities, CEA and 80,000 Hours can be much more effective than they would by trying to do everything.
I understand this is a lower bound, but these numbers seem embarrassingly low given how generous the match was, e.g. compared to the old Facebook giving Tuesday match results. I guess most people don't read this forum anymore and we should have instead shared this and other important topics on Twitter?
I'd recommend asking these questions to your favourite AI or to the facilitators, my understanding is that course facilitators don't use this forum for various reasons, and all of ChatGPT, Claude, and Gemini are great at answering these questions. E.g. here's ChatGPT:
I think you’ve identified a real jump in context, rather than missed some prerequisite reading!
This was originally a post on LessWrong.com, an online community focused on reasoning and decision-making. It does two things: first, illustrates how cognitive biases can feel like reasonable thinking from the inside; second, argues that these concrete illustrations are a better way to teach people about biases than abstract definitions. That second part addresses the original community’s communication habits, which explains the references to “LessWrongians” and makes it feel like a tangent here.
The connection to effective altruism is that trying to help effectively requires assessing our own judgments critically. For example, “I’ve carefully researched this charity and everything supports donating to it” might reflect good research—or selectively seeking favourable evidence. The tables are meant to help us notice that possibility, much as the scout mindset encourages us to examine beliefs we’re attached to. They aren’t proof that any particular belief is biased.
So your interpretation seems right: the main takeaway for this chapter is awareness of cognitive biases. The discussion of communication explains why the author presented them this way. An introductory note distinguishing those two purposes would make its inclusion much clearer.
it seems like cash transfers that improve annual income by 10-30% in the USA have pretty much no effect on wellbeing. Weird. Not sure why yet.
I don't think it's that weird with the above values of η, considering that even in low income countries the effects of cash are significant but not immense, see this previous thread
we fail to appreciate why everyone is so mad at us.
I really don't think FTX is the main reason why everyone is so mad at you (read e.g. The Economist article mentioned, or search for "effective altruism" online), I think this should be taken as an opportunity to clarify other common misconceptions, instead of taking even more blame for being associated with some of the organizations that praised SBF for donating a lot of money to charity or advised him on how to donate it
I'm sad to see that it's now practically unusuable for many (EA) charities and donors.
I don't think this is true for that many. As far as I can see most EA charities and donors privately declined to participate in this pressure campaign against Manifund (and because of that you don't see them posting here), and the ~6 that did participate as far as I know are not funding constrained (edit: and/or have easier ways of receiving donations without the 5% fee), so wouldn't have had much use for Manifund in the first place.
I used to think something along those lines, but given what happened since the release of Opus 4.5 in November 2025, I now personally believe that EA as a whole is not focusing on AI enough.
I also believe that 80,000 Hours doesn't get enough credit for having pushed for focusing on what's most important since well before it was so clear and popular, and people like me who were complaining about them just telling everyone to work in AI should apologize and update sooner rather than later.
Some thoughts on your points:
I agree that 80k was possibly the best, but the two you link are also good, there's also https://forum.effectivealtruism.org/handbook which imho is still decent despite (or maybe thanks to?) being unmaintained
I personally strongly disagree with this, EA should be about helping others effectively. “the science of doing good” should reject "alternative science".
I'm not claiming that there is consensus among reasonable people that trying to influence the potential upcoming transition in a positive way is the best way to help others, but you shouldn't embrace diversity of views for its own sake.
This seems less important than people moving into careers that are better for the world, or donating more money, or influencing policy, or many other things. So I think it makes sense that more and more organizations are focusing on results rather than on EA's growth. I also don't think it's true (the US Department of War tweeted about effective altruism, The Economist had an article on EA on its front page, there's a major Netflix series coming out in November, ...)
Also, lots of people are still getting into non-AI EA from 80k. In the past 6 months 24 people signed the 10% pledge with 80k as a primary source. Given the massive interest in AI and 80k's AI focus, I expect that tons of people are discovering 80k and through it EA.
I think CEA is focusing on having very simplified, normie-friendly pages, without mentioning anything weird like the fact that most life in the next century might be of non-biological origin and deeply uncertain about its own moral status. I think it makes sense for CEA's content to focus on the EA brand and for 80k on how to do the most good with your career (see below on focus)
I think there's a very important difference between not endorsing those ideas because you disagree with them and not endorsing them because they're too weird or far from the social norm. I also feel uncomfortable mentioning significant donations and animal welfare, but I think that positively correlates with those things being very important to talk more about, because I deeply believe the related ideas to be true, and the discomfort is due to mistaken norms and beliefs.
Those are very far from the movement's most bizarre ideas, they are all reasonable things to focus on, and 80k's reasoning for focusing on those is well laid out, they're obviously not writing them just because they are bizarre.
To be clear I think it would be weird and bad if CEA redefined "effective altruism" to be about positively influencing this potential transition, instead of being about trying to help others from a broad range of perspectives, but I think what 80k is doing is very good for EA and for the world, and many more of us should be doing similar things.
I personally don't think you need to commit to the view that AI takeover is likely to encourage more focus on influencing upcoming transformative technology, you just need to believe that it's reasonably likely to be transformative and that there are reasonable enough theories of change to try to influence that transition in a more positive way.
Separately from considerations about the relative importance of AI, I also think from the outside it can be very hard to appreciate the value of focusing (see e.g. GWWC retiring 10 initiatives, which internally we called "Project Focus", led to clear significant growth in our key metrics and was extremely successful).
By focusing more on their respective priorities, CEA and 80,000 Hours can be much more effective than they would by trying to do everything.
For reference, Facebook's giving Tuesday match mentioned above lasted seconds https://en.wikipedia.org/wiki/GivingTuesday#History
I understand this is a lower bound, but these numbers seem embarrassingly low given how generous the match was, e.g. compared to the old Facebook giving Tuesday match results. I guess most people don't read this forum anymore and we should have instead shared this and other important topics on Twitter?
Old numbers for reference
Are you going through the CEA Intro Program by any chance? https://www.effectivealtruism.org/courses/introductory-program
I'd recommend asking these questions to your favourite AI or to the facilitators, my understanding is that course facilitators don't use this forum for various reasons, and all of ChatGPT, Claude, and Gemini are great at answering these questions. E.g. here's ChatGPT:
Thanks so much for this post!
I know much less about this than you do, but some quick thoughts:
Is that really the convention? I thought the convention was to assume https://en.wikipedia.org/wiki/Isoelastic_utility with a η between 1.2 and 1.5?
I don't think it's that weird with the above values of η, considering that even in low income countries the effects of cash are significant but not immense, see this previous thread
I really don't think FTX is the main reason why everyone is so mad at you (read e.g. The Economist article mentioned, or search for "effective altruism" online), I think this should be taken as an opportunity to clarify other common misconceptions, instead of taking even more blame for being associated with some of the organizations that praised SBF for donating a lot of money to charity or advised him on how to donate it
You might be interested in Most* small probabilities aren't pascalian , I think a better example is looking both ways before crossing the street
Giving What We Can also did not receive anything close to $1.2 billion in donations in 2024 or $2 billion in 2025, that's the total money moved by the whole ~EA-ish effective giving ecosystem (mostly Coefficient Giving and GiveWell), you can read more here https://forum.effectivealtruism.org/posts/T7CKswFYSgjePAwZh/updates-on-the-effective-giving-ecosystem-2026
The "$15 Billion per year" seems to be from Wired and based on Ransohoff's estimate of ~$150B in pledged amounts from Anthropic times ~10%
I don't think this is true for that many. As far as I can see most EA charities and donors privately declined to participate in this pressure campaign against Manifund (and because of that you don't see them posting here), and the ~6 that did participate as far as I know are not funding constrained (edit: and/or have easier ways of receiving donations without the 5% fee), so wouldn't have had much use for Manifund in the first place.