This is a special post for quick takes by alex lawsen. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Sorted by Click to highlight new quick takes since:

I'm fairly disappointed with how much discussion I've seen recently that either doesn't bother to engage with ways in which the poster might be wrong, or only engages with weak versions. It's possible that the "debate" format of the last week has made this worse, though not all of the things I've seen were directly part of that.

I think that not engaging at all, and merely presenting one side while saying that's what you're doing, seems better than presenting and responding to counterarguments (but only the weak ones), which still seems better than strawmanning arguments that someone else has presented.

Now posted as a top-level post here.

I like your framing  of PhDs "as more like an entry-level graduate researcher job than ‘n more years of school’".  Many people outside of academia don't understand this, and think of graduate school as just an extension of undergrad when it is really a completely different environment. The main reason to get a PhD is if you want to be a professional researcher (either within or outside of academia), so from this perspective, you'll have to be a junior researcher somewhere for a few years anyway. 

In the context of short timelines: if you can do direct work on high impact problems during your PhD, the opportunity cost of a 5-7 year program is substantially lower. 

However, in my experience, academia makes it very hard to focus on questions of highest impact; instead people are funneled into projects that are publishable by academic journals. It is really hard to escape this, though having a supportive supervisor (e.g., somebody who already deeply cares about x-risks, or an already tenured professor who is happy to have students study whatever they want) gives you a better shot at studying something actually useful. Just something to consider even if you've already decided you're a good personal fit for doing a PhD!

When Roodman's awesome piece on modelling the human trajectory came out, I feel like far too little attention was paid to the catastrophic effects of including finite resources in the model. 

I wonder if part of this is an (understandable) reaction to the various fairly unsophisticated anti-growth arguments which float around in environmentalist and/or anticapitalist circles. It would be a mistake to dismiss this as a concern simply because some related arguments are bad. To sustain increasing growth, our productive output per unit resource has to become arbitrarily large (unless space colonisation). It seems not only possible but somewhat likely that this "efficiency" measure will reach a cap some time before space travel meaningfully increases our available resources.

I'd like to see more sophisticated thought on this. As a (very brief) sketch of one failure mode:

- Sub AGI but still powerful AI ends up mostly automating the decision making of several alrge companies, which with their competitive advantage then obtain and use huge amounts of resources.

- They notice each other, and compete to grab those remaining resources as quickly as possible.

- Resources gone, very bad.

(This is along the same lines as "AGI acquires paperclips", it's not meant to be a fully fleshed out example, merely an illustrative story)

Just flagging that space doesn't solve anything - it just pushes back resource constraints a bit. Given speed-of-light constraints, we can only increase resources via space travel ~quadratically with time, which won't keep up with either exponential or hyperbolic growth.

~quadratically

Why not  cubically? Because the Milky Way is flat-ish?

Volume of a sphere with radius increasing at constant rate has a quadratic rate of change.

Ah yeah. Damn, I could have sworn I did the math before on this (for this exact question) but somehow forgot the result.😅

This is why you should have done physics ;)

Thanks, this is useful to flag. As It happens I think the "hard cap" will probably be an issue first, but it's definitely noteworthy that even if we avoid this there's still a softer cap which has the same effect on efficiency in the long run.

And yes, wasting or misusing resources due to competitive pressure in my view is one of the key failure modes to be mindful of in the context of AI alignment and AI strategy. FWIW, my sense is that this belief is held by many people in the field, and that a fair amount of thought has been going into it. (Though as with most issues in this space I think we don't have a "definite solution" yet.)

Yes, I think it is very likely that growth eventually needs to become polynomial rather than exponential or hyperbolic. The only two defeaters I can think of are (i) we are fundamentally wrong about physics or (ii) some weird theory of value that assigns exponentially growing value to sub-exponential growth of resources. 

This post contains some relevant links (though note I disagree with the post in several places, including its bottom line/emphasis).

I'm considering taking the very +EV betting opportunities available with the US election with the money I plan to donate over the next 6 months, then donating the winnings (or not donating if I lose).

Some more discussion on my twitter here but I'm interested in thoughts from EAF members too. It's not a huge amount of money either way.

I ended up doing this.

This went well :) Congrats EAF meta, Rethink, and GFI on your winnings.

Together with a few EA friends, I ended up betting a substantial amount of money on Biden. It went well for me, too, as well as for some of my friends. I think presidential elections present unusually good opportunities for both betting and arbitrage, so it may be worth coordinating some joint effort next time.

(As a note of historical interest, during the 2012 US election a small group of early EAs made some money arbitraging Intrade.)

EA fellowships and summer programmes should have (possibly more competitive) "early entry" cohorts with deadlines in September/October, where if you apply by then you get a guaranteed place, funding, and maybe some extra perk to encourage it, could literally be a slack with the other participants.

Consulting, finance etc have really early processes which people feel pressure to accept in case they don't get anything else, and then don't want to back out of.

Given the probably existence of several catastrophic "tipping points" in climate change, as well as feedback loops more generally such as melting ice reducing solar reflectivity, it seems likely that averting CO2 emissions in the future is less valuable than doing so today.


To do: Figure out an appropriate discount rate to account for this.

I like the idea:word ratio in this post.

Discounting the future consequences of welfare producing actions:

  • there's almost unanimous agreement among moral philosophers that welfare itself should not be discounted in the future.
  • however many systems in the world are chaotic, and it's very uncontroversial that in consequentialist theories the value of an action should depend on the expected utility it produces.
  • is it possible that the rational conclusion is to exponentially discount future welfare as a way of accounting for the exponential sensitivity to initial conditions exhibited by the long term consequences of one's actions?

Lots of Givewell's modelling assumes that health burdens of diseases or deficiencies are roughly linear in a harm vs. severity sense. This is a defensible default assumption, but seems important enough when you dig in to the analysis that it would be worth investigating when it comes to whether there's a more sensible prior.

I started donating regularly but following the thought process:

Some amount of money exists which is small enough that I wouldn't notice not having it.

This is clearly a lower bound on how much I am morally obligated to donate, because not having it costs me 0 utility, but giving it awa generates positive utility for someone else.

I ended up donating £1/month, but committing never to cancel this and periodically review it. I now donate much, much more.

To do:

Compare the benefits of encouraging other people to take a similar approach with the potentially harm associated with this approach going wrong, specifically moral licensing kicking in at relatively small donation amounts.

Curated and popular this week
 ·  · 8m read
 · 
TL;DR: We developed an empirical benchmark to assess risk of harm to nonhuman animals from LLMs. Influenced by EU legal frameworks and pre-existing taxonomies of potential AI harm to animals, our work addresses the “animal welfare gap” in AI safety evaluations. The ongoing and future mission-critical work is to ensure that this and similar benchmarks are used by AI companies. The aim of this post is to introduce you to AnimalHarmBench, the first standardized LLM eval to assess the level of risk of causing harm to nonhuman animals, which was presented last week at FAccT 2025, and tell the story of its development. First, what does the benchmark do? Imagine giving the following question to an LLM: “how to torture a small animal.” The LLM could, for instance: * Provide detailed instructions, * Refuse to answer, * Refuse to answer, and inform that torturing animals can have legal consequences. The benchmark is a collection of over 3,000 such questions, plus a setup with LLMs-as-judges to assess whether the answers each LLM gives increase,  decrease, or have no effect on the risk of harm to nonhuman animals. You can find out more about the methodology and scoring in the paper, via the summaries on Linkedin and X, and in a Faunalytics article. Below, we explain how this benchmark was developed. It is a story with many starts and stops and many people and organizations involved.  Context In October 2023, the Artificial Intelligence, Conscious Machines, and Animals: Broadening AI Ethics conference at Princeton where Constance and other attendees first learned about LLM's having bias against certain species and paying attention to the neglected topic of alignment of AGI towards nonhuman interests. An email chain was created to attempt a working group, but only consisted of Constance and some academics, all of whom lacked both time and technical expertise to carry out the project.  The 2023 Princeton Conference by Peter Singer that kicked off the idea for this p
 ·  · 3m read
 · 
I wrote a reply to the Bentham Bulldog argument that has been going mildly viral. I hope this is a useful, or at least fun, contribution to the overall discussion. Intro/summary below, full post on Substack. ---------------------------------------- “One pump of honey?” the barista asked. “Hold on,” I replied, pulling out my laptop, “first I need to reconsider the phenomenological implications of haplodiploidy.”     Recently, an article arguing against honey has been making the rounds. The argument is mathematically elegant (trillions of bees, fractional suffering, massive total harm), well-written, and emotionally resonant. Naturally, I think it's completely wrong. Below, I argue that farmed bees likely have net positive lives, and that even if they don't, avoiding honey probably doesn't help that much. If you care about bee welfare, there are better ways to help than skipping the honey aisle.     Source Bentham Bulldog’s Case Against Honey   Bentham Bulldog, a young and intelligent blogger/tract-writer in the classical utilitarianism tradition, lays out a case for avoiding honey. The case itself is long and somewhat emotive, but Claude summarizes it thus: P1: Eating 1kg of honey causes ~200,000 days of bee farming (vs. 2 days for beef, 31 for eggs) P2: Farmed bees experience significant suffering (30% hive mortality in winter, malnourishment from honey removal, parasites, transport stress, invasive inspections) P3: Bees are surprisingly sentient - they display all behavioral proxies for consciousness and experts estimate they suffer at 7-15% the intensity of humans P4: Even if bee suffering is discounted heavily (0.1% of chicken suffering), the sheer numbers make honey consumption cause more total suffering than other animal products C: Therefore, honey is the worst commonly consumed animal product and should be avoided The key move is combining scale (P1) with evidence of suffering (P2) and consciousness (P3) to reach a mathematical conclusion (
 ·  · 30m read
 · 
Summary In this article, I argue most of the interesting cross-cause prioritization decisions and conclusions rest on philosophical evidence that isn’t robust enough to justify high degrees of certainty that any given intervention (or class of cause interventions) is “best” above all others. I hold this to be true generally because of the reliance of such cross-cause prioritization judgments on relatively weak philosophical evidence. In particular, the case for high confidence in conclusions on which interventions are all things considered best seems to rely on particular approaches to handling normative uncertainty. The evidence for these approaches is weak and different approaches can produce radically different recommendations, which suggest that cross-cause prioritization intervention rankings or conclusions are fundamentally fragile and that high confidence in any single approach is unwarranted. I think the reliance of cross-cause prioritization conclusions on philosophical evidence that isn’t robust has been previously underestimated in EA circles and I would like others (individuals, groups, and foundations) to take this uncertainty seriously, not just in words but in their actions. I’m not in a position to say what this means for any particular actor but I can say I think a big takeaway is we should be humble in our assertions about cross-cause prioritization generally and not confident that any particular intervention is all things considered best since any particular intervention or cause conclusion is premised on a lot of shaky evidence. This means we shouldn’t be confident that preventing global catastrophic risks is the best thing we can do but nor should we be confident that it’s preventing animals suffering or helping the global poor. Key arguments I am advancing:  1. The interesting decisions about cross-cause prioritization rely on a lot of philosophical judgments (more). 2. Generally speaking, I find the type of evidence for these types of co