David T

Karma: 905

David T 17 Dec 2024 21:22 UTC
3 points
0 ∶ 1
in reply to: Jason’s comment on: My Problem with Veganism and Deontology
The other problem with the “indirect enough” argument is that the donations are even more indirect
Sure, the meat people eat is usually killed long before it’s ordered and eating a few dozen chickens per year doesn’t individually shift an industry. But likewise, a $1000 donation doesn’t meaningfully affect an advocacy charity’s ability to win a court case.
Both only work in aggregate, and on a causal basis the link between meat demand and factory farming is much more robustly-established than advocacy charity income and relative absence of factory farms^[1]
And standards for crediting impact need to be stricter here because multiple counting is a much more meaningful problem than when an altruistic donor is deciding where to donate
This is a good point too. If you’re using donations to prioritise in a counterfactual scenario, what part of the outcome is actually “your impact” is irrelevant. If you’re using them to buy indulgences, that’s less obviously the case.
1. ^
  on a money basis it’s less certain, but I still don’t think vegan diets are dramatically more expensive than meat ones, and the DALY impact of eating half a chicken doesn’t seem to be very different from favourable estimates of DALY impact of a dollar donation to Legal Impact for Chickens…

David T 15 Dec 2024 14:47 UTC
9 points
5 ∶ 0
on: “Charity” as a conflationary alliance term
A large part of backlash against effective altruism comes from people worried about EA ideals being corrosive to the “paying for public goods” or “partial philanthropy” mechanisms.
I think this is a good point. I don’t think it’s a particularly strong argument against EA, not least because EA doesn’t appear to be having any discernible impact on people’s willingness to fund climbing organizations or conference halls, but it certainly comes up a lot in critical articles.
More common forms of ostensibly “impartial” giving, like supporting global health initiatives or animal welfare, are probably better understood as examples of partial philanthropy with extended notions of “we”, like “we, living humans” or “we, mammals”.
But I don’t agree with this. Giving anonymously to unknown recipients in a faraway country via an unconnected small NGO doesn’t have any of the typical benefits associated with supporting a “collective we” (anticipated reciprocity, kin selection, identity, chauvinism against perceived enemies etc) making it about as impartial as it gets, and I don’t think people care about chicken welfare out of collective identity, never mind a stronger sense of collective identity than with potential future humans. Indeed it would be far easier to class many longtermist organizations under your definition of “partial philanthropy” as recipients are typically known members of a community with shared beliefs (and sometimes social circles), and the immediate benefit is often research the donor and the donor’s community find particularly interesting.
I think the factors you’ve highlighted that apply to some types of charity like access to public goods overlap with other motivations for giving like sense of duty, feelings of satisfaction and signalling which apply to all types of charity.

David T 13 Dec 2024 23:36 UTC
3 points
0 ∶ 0
in reply to: Lukas_Gloor’s comment on: Probabilities might be off by one percentage point
Thanks for the thoughtful response.
On (1) I’m not really sure the uncertainty and the trust in the estimate are separable. A probability estimate of a nonrecurring event^[1] fundamentally is a label someone^[2] applies to how confident they are something will happen. A corollary of this is that you should probably take into account how probability estimates could have actually been reached, your trust in that reasoning and the likelihood of bias when deciding how to act. ^[3]
On (2) I agree with your comments about the OP’s point; if the probabilities are +/-1 percentage point with error symmetrically distributed they’re still on average 1.5%^[4], though in some circumstances introducing error bars might affect how you handle risk. But as I’ve said, I don’t think the distribution of errors looks like this when it comes to assessing whether long shots are worth pursuing or not (not even under the assumption of good faith). I’d be pretty worried if hits based grant-makers didn’t, frankly, and this question puts me in their shoes.
Your point about analytic philosophy often expecting literal answers to slightly weird hypotheticals is a good one. But EA isn’t just analytic philosophy and St Petersburg Paradoxes, it’s also people literally coming up with best guesses of probabilities of things they think might work and multiplying them (and a whole subculture based on that, and guesstimating just how impactful “crazy train” long shot ideas they’re curious about might be). So I think it’s pretty reasonable to treat it not as a slightly daft hypothetical where a 1.5% probability is an empirical reality,^[5] but as a real world decision grant award scenario where the “1.5% probability” is a suspiciously precise credence, and you’ve got to decide whether to trust it enough to fund it over something that definitely works. In that situation, I think I’m discounting the estimated chance of success of the long shot by more than 50%.
FWIW I don’t take the question as evidence the survey designers are biased in any way
1. ^
  “this will either avert 100,000 DALYs or have no effect” doesn’t feel like a proposition based on well-evidenced statistical regularities...
2. ^
  not me. Or at least a “1.5%” chance of working for thousands of people and implicitly a 98.5% chance of having no effect on anyone certainly doesn’t feel like the sort of degree of precision I’d estimate to...
3. ^
  Whilst it’s the unintended consequences of how the question was framed, this example feels particularly fishy. We’re asked to contemplate trading off something that certainly will work against something potentially higher yielding that is highly unlikely to work, and yet the thing that is highly unlikely to work turns out to have the higher EV because someone has speculated on its likelihood to a very high degree of precision, and those extra 5 thousandths made all the difference. What’s the chance the latter estimate is completely bogus or finessed to favour the latter option? I’d say in real world scenarios (and certainly not just EA scenarios) it’s quite a bit more than 5 in 1000....
4. ^
  that one’s a math test too ;-)
5. ^
  maybe a universe where physics is a god with an RNG...

David T 12 Dec 2024 23:35 UTC
1 point
0 ∶ 0
in reply to: Lukas_Gloor’s comment on: Probabilities might be off by one percentage point
Feels like taking into account the likelihood that the “1.5% probability of 100,000 DALYs averted” estimate is a credence based on some marginally-relevant base rate^[1] that might have been chosen with a significant bias towards optimism is very much in keeping with the spirit of the question (which presumably is about gauging attitudes towards uncertainty, not testing basic EV calculation skills)^[2].
A very low percentage chance of averting a lot of DALYs feels a lot more like “1.5% of clinical trials of therapies for X succeeded; this untested idea might also have a 1.5% chance” optimism attached to a proposal offering little reason to believe it’s above average rather than an estimate based on somewhat robust statistics (we inferred that 1.5% of people who receive this drug will be cured from the 1.5% of people who had that outcome in trials). So it seems quite reasonable to assume that the 1.5% chance of a positive binary outcome estimate might be biased upwards. Even more so in the context of “we acknowledge this is a long shot and high-certainty solutions to other pressing problems exist, but if the chance of this making an impact was as high as 0.0x%...” style fundraising appeals to EAs’ determination to avoid scope insensitivity.
1. ^
  either that or someone’s been remarkably precise in their subjective estimates or collected some unusual type of empirical data. I certainly can’t imagine reaching the conclusion an option has exactly 1.5% chance of averting 100k DALYs myself
2. ^
  if you want to show off you understand EV and risk estimation you’d answer (C) “here’s how I’d construct my portfolio” anyway :-)

David T 12 Dec 2024 22:12 UTC
2 points
4 ∶ 0
in reply to: Habryka’s comment on: Upcoming changes to Open Philanthropy’s university group funding
They do specifically say that they consider other types of university funding to have greater cost-benefit (and I don’t think it makes sense to exclude reputation concerns from cost-benefit analysis, particularly when reputation boost is a large part of the benefit being paid for in the first place). Presumably not paying stipends would leave more to go around. I agree that more detail would be welcome.

David T 12 Dec 2024 21:57 UTC
1 point
0 ∶ 0
on: “Charity” as a conflationary alliance term
A large part of backlash against effective altruism comes from people worried about EA ideals being corrosive to the “paying for public goods” or “partial philanthropy” mechanisms.
I think this is a good observation (I think the worry highlighted is one of the weakest arguments against EA, not least because EA has very limited real world impact upon the amount spent on animal shelters or concert halls, but it definitely comes up a lot in articles people like sharing on here).
More common forms of ostensibly “impartial” giving, like supporting global health initiatives or animal welfare, are probably better understood as examples of partial philanthropy with extended notions of “we”, like “we, living humans” or “we, mammals”.
I don’t agree with this though. I don’t think people donate to anonymous poor recipients in faraway countries or farm animals out of sense of collective identity. There’s little or no reciprocal altruism or collective identity there (particularly when it comes to the animals). I don’t think donating to exploring ideas of future people or digital minds is more impartial simply because these don’t [yet] exist. (Indeed I think it would be easier to characterise some of the niche longtermist research donations as being “partial” philanthropy on the basis that the recipients are typically known and respected members of an established in-group, with shared [unusual] interests, and the outcome is often research whose most obviously quantifiable impact is that the donor and their group find it very interesting. That strikes me as similar to quite a lot of other philanthropic research funding including in academia).
I think the “types” of charity are better understood as a set of motivations which overlap (and also include others like fuzzy feelings of satisfaction, signalling, interests, sense of duty etc which can coexist with also being a user of that conference hall or a fellow Christian or someone that believes its important future humanity ). Donating to AMF is about as impartial as it gets in terms of outcome, but there’s definitely some sort of collective identity benefit to doing so whilst identifying as part of a group with a shared epistemology and understanding that points towards donating mattering and outcomes mattering and AMF being a good way to achieve this. Ditto impartial donations to mainstream charity made by people who have a sense of religious duty to random strangers, or completely impartial donations to a research funding pool made by people with strong convictions about progress.

David T 12 Dec 2024 21:06 UTC
6 points
1 ∶ 0
in reply to: Habryka’s comment on: Upcoming changes to Open Philanthropy’s university group funding
I think it’s fair enough to caution against purely performative frugality. But I’m not sure the OP even justifies the suggestion that the organizers actually are more cost effective (they concluded the difference between paid and unpaid organizers’ individual contributions were “substantive, not enormous”; there’s a difference between paid people doing more work than volunteers and it being more cost effective to pay...). That’s even more the case if you take into account that the primary role of an effective university organizer is attracting more people (or “low context observers”) to become more altruistic and this instance of the “weirdness” argument is essentially that paying students undercut the group’s ability to appeal to people on altruistic grounds, even if individual paid staff put in more effort. And they were unusually well paid by campus standards for tasks almost every other student society use volunteers for.^[1] And that there’s no evidence that the other ways CEA proposes spending the money instead are less effective.
1. ^
  one area we might agree is that I’m not sure if OpenPhil considered alternatives like making stipends needs-based or just a bit lower and more focused as a pragmatic alternative to just cancelling them altogether.

David T 12 Dec 2024 20:33 UTC
1 point
0 ∶ 0
in reply to: David T’s comment on: NYT—What if Charity Shouldn’t be Optimized
Perplexed by the reaction here. Not sure what people are taking most issue with:.
Me saying the stereotypes were limited and not particularly negative? If you think a reference to being disproportionately funded by a small number of tech billionaires, (balanced out by also accurate references to Singer and the prior emergence of a movement and an example of Berger giving a kidney rather than money) is negative stereotyping, you haven’t read other critical takes on EA, never mind experienced what some other “minorities” deal with on a daily basis!
Me saying the more informed and effective criticisms of EA and EA orgs tended to point out where they fall well short of the rigour they demand? Again, I’d have thought it was glaringly obvious, whether it’s nuanced insider criticism of specific inconsistencies in outcome measures or reviews of specific organizations, or drive-by observations that buying Wytham Abbey or early-stage funding for OpenAI may not have been high points of evidence-based philanthropy. That’s obviously more useful than “these people have a different worldview” type articles like this. Even some of the purely stereotype-based criticisms of the money sloshing around the FTX ecosystem probably weren’t “stopped clock” moments...
Or me pointing out that EAs also criticise non EAs’ philanthropic choices, sometimes in generic terms? If you haven’t read Peter Singer writing how other people have the wrong philanthropic priorities, you haven’t read much Peter Singer!

David T 11 Dec 2024 20:21 UTC
1 point
0 ∶ 0
in reply to: Greg_Colbourn’s comment on: OpenAI’s o1 tried to avoid being shut down, and lied about it, in evals
Its not role playing if the AI behaves completely differently except when the instruction set is specifically set up so the AI parses it as role playing, in this case with other glaringly obviously unrealistic scene-setting like sending it a config file with it’s own monitoring settings along with a logic puzzle involving a concealed goal. I can get a bucket of water to conduct a sneak attack on me if I put enough effort into setting it up to fall on my head, but that doesn’t mean I need to worry about being ambushed by water on a regular basis!

David T 8 Dec 2024 22:09 UTC
−7 points
1 ∶ 3
in reply to: Dicentra’s comment on: NYT—What if Charity Shouldn’t be Optimized
I don’t think they would for people with unusual hobbies or lifestyle choices or belief sets, with stereotypes related to those things.
And the “stereotyping” in here is really limited and not particularly negative: there’s space apportioned to highlighting how OpenPhil’s chief executive gave a kidney for the cause and none to stereotypes of WEIRD Bay Area nerds or Oxford ivory towers or effective partying in the Bahamas. If you knew nothing else about the movement, you’d probably come away with the conclusion that EAs were a bit too consistent in obsessing over measurable outcomes; most of the more informed and effective criticisms argue the opposite!
(It also ends up by suggesting that EA as a philosophy offers a set of questions that are worth asking and some of its typical answers are perfectly valid. Think most minorities would love it if outside criticism of their culture generally drew that sort of conclusion!)
EAs can and do write opinion pieces broadly or specifically criticising other people’s philanthropic choices all the time. I don’t think EA should be exempted from such arguments.

David T 8 Dec 2024 22:04 UTC
7 points
1 ∶ 0
in reply to: Joseph Lemien’s comment on: NYT—What if Charity Shouldn’t be Optimized
So it stuck me as odd to have such an established set of solutions which weren’t even mentioned. How to Measure Anything is great, but there is also lots written about willingness to pay.
I agree the article isn’t particularly deep, but the plurality of possible measures arguably supports the central argument which appears to be that EA approaches to quantifying philanthropy isn’t the be all and end all.^[1] Willingness to pay, for example, is a measure which works against arguments by Singer that money voluntarily donated to the Notre Dame roof would be better redirected to alleviating global suffering.
1. ^
  wait until she discovers how differently some EAs quantify different types of intervention!

David T 8 Dec 2024 20:37 UTC
2 points
0 ∶ 0
in reply to: huw’s comment on: NYT—What if Charity Shouldn’t be Optimized
It’s notable, of course, that the article mentioned the Notre Dame, and not the ongoing destruction of religious history in Gaza or Syria or Afghanistan or Sudan or Ukraine (for example
Not really. Notre Dame was mentioned because some prominent EAs have criticised its expensive restoration project as being an inappropriate use of philanthropic funding. As far as I’m aware, prominent EAs haven’t devoted the same criticism to the opulence of Hindu or Buddhist monuments or attempts to protect antiquities in conflict zones, and I don’t think that makes them racist or anglo-centric either.
Now people can and do make arguments for preserving archaeological sites in poorer countries on the grounds of them being more vulnerable and less expensive to repair which is essentially a cost-effectiveness argument. No doubt they would agree with your suggestion to direct giving outward, but I don’t think that group overlaps with EAs at all. (And for those who think that rebuilding destroyed historical sites are a valid use of philanthropic funding there are also obvious arguments that people reasonably prefer to donate to things that they can see and that can be enjoyed by millions of people over, say, the restoration of the Bamiyan Buddhas in a remote area of a wartorn country since taken over by the entity which originally intentionally destroyed them. Nevertheless, there were serious discussions about restoring the Bamiyan Buddhas prior to the Taliban resurgence, but I don’t think EA had anything to do with any of the debate)

David T 7 Dec 2024 16:37 UTC
4 points
1 ∶ 1
in reply to: Greg_Colbourn’s comment on: OpenAI’s o1 tried to avoid being shut down, and lied about it, in evals
I didn’t downvote, but I’ll offer an alternative reading of the experiment discussed here some people unimpressed by the arguments may agree with, especially if they’ve read the paper.
Some LLMs were given some roleplaying prompts to solve a problem intentionally constructed to make it impossible for them to behave in an “aligned” manner, where arguably “subversion” was the more ethical course of action.
In this role-playing context, the models very occasionally generated responses which solved a problem broadly of the form “you need to do y, but people who want to stop you from doing y are watching you on camera. On their camera is a switch”. On some of the occasions they solved that initial problem and were greeted by a followup role play interrogating them about their subterfuge, they sometimes chose denial (and sometimes output chain of thought arguments for why denial was better than confession), and on a small fraction of those denials actually successfully continued that chain of thought to provide false answers [coherent or otherwise] for multiple consecutive followups.
To put it mildly, this doesn’t sound like “scheming” or converging on self preservation as an instrumental goal, it sounds like when a model is prompted that it’s role-playing an entity which must prioritise something over subsequent instructions, it occasionally finds the deliberately-constructed only valid route to achieving that priority, which in this case happens to have been one involving self preservation intended to trigger AI doomers. Especially considering how many prompts they burned through to generate the scary sounding answers and how many times the AI generated responses inconsistent with self-preservation.
The models shouldn’t be given control of important stuff because their responses are unpredictable and a bit stupid, not because they’re exhibiting signs of emergent agency.

David T 3 Dec 2024 11:39 UTC
2 points
0 ∶ 0
in reply to: ivn’s comment on: Is it an Exceptionally Good Time for a “Third Major Political Party” in the US?
The Democrats absorbed some of the disaffected Republicans and are the natural party of anyone who just really doesn’t want Trump, and the Libertarians are closer to what many disaffected Republicans believe, but have obviously failed to take advantage of it.

Trump’s turnout and approval ratings on the right are really high, which doesn’t exactly suggest there’s scope for another mass membership party there

David T 3 Dec 2024 11:12 UTC
1 point
1 ∶ 0
on: Is it an Exceptionally Good Time for a “Third Major Political Party” in the US?
What is the “third major political party” supposed to represent and why would anyone vote for them?

There’s already two parties for disaffected Republicans, one of which already has a natural support base and came second in the election. Turns out most of them weren’t that disaffected though...

David T 30 Nov 2024 18:51 UTC
3 points
2 ∶ 0
in reply to: Toby Tremlett🔹’s comment on: Can I put EA on my resume?
I’ll further add that most people aren’t going to bother doing the quick Google, they’re going to see “organised university society” and whatever two sentence summary you’ve got about it being charity-related and see it as a positive, although not necessarily any more positive than organising the RAG week charity or a sports team.
The bigger question is if and how you raise it as an answer to a question about your life experiences at interview
(FWIW my ad-blocked Google results for Effective Altruism are this website, the Wikipedia link and a BBC article about SBF)

David T 30 Nov 2024 11:24 UTC
1 point
0 ∶ 0
in reply to: Thomas Kwa’s comment on: Thomas Kwa’s Shortform
Are you saying that people were worried about abolition slowing down economic growth and lowering standards of living? I haven’t heard this as a significant concern—free labor was perfectly capable of producing cotton at a small premium, and there were significant British boycotts of slave-produced products like cotton and sugar.
Absolutely slaveholders and those dependent on them were worried about their own standard of living (and more importantly, specifically not interested in significantly improving the standard of living of plantation slaves, and not because they’d never heard anyone put forward the idea that all people were equal. I mean, some of them were on first name terms with Thomas Paine and signed the Declaration of Independence and still didn’t release their slaves!). I’m sure most people who were sympathetic to EA ideas would have strongly disagreed with this prioritisation decision, just like the Quakers or Jeremy Bentham. I just don’t think they’d have been more influential than the Quakers or Jeremy Bentham, or indeed the deeply religious abolitionists lead by William Wilberforce.
I agree the number of philosophers in EA is quite low, but I’m assuming the influence centre would be similar, possibly even more Oxford-centric, in a pre-internet, status-obsessed social environment where discourse is more centred on physical places and elite institutions^[1]. For related reasons I think they’d be influential in the sort of place where abolitionist arguments were already getting a fair hearing, and of little consequence in slaveowning towns in the Deep South. In the UK, I think the political process was held up by the amount of vested interests in keeping it going in Parliament and beliefs that slavery was “the natural order” rather than any lack of zeal or arguments or resources on the abolitionist side (though I’m sure they’d have been grateful for press baron Moskovitz’s donations!). I think you could make the argument that slave trade abolition in the UK was actually pretty early considering the revenues it generated, who benefited, and the generally deeply inegalitarian social values and assumption of racial superiority of British society at the time.
As for utilitarian arguments, that’s not the main way I imagine EAs would help. EA pragmatists would prioritize the cause for utilitarian reasons and do whatever is best to achieve their policy goals, much as we are already doing for animal welfare. The success of EAs in animal welfare, or indeed anywhere other than x-risk, is in implementation of things like corporate campaigns rather than mass spreading of arguments.
I agree this is probably the main way that EAs would try to help, I just don’t think abolitionism is an area where this would have much impact (both due to it being a centralised political problem, and due to existing abolitionists already being quite sophisticated when it came to boycotts and literature distribution and freedom suits). I can totally imagine historic EAs trying to find more efficient means to distribute poor relief and shaming bad farmers (and doing some stuff that seemed like a it might be effective at the time like abstinence campaigns to ward off Mathusian catastrophes), I just don’t think it would change much on the politics of slavery which wars were to be fought over. I guess it also depends on how much knowledge they get to take back with them. Introducing RCTs to medicine in the 1770s would have been revolutionary!^[2]
I agree a more general “what would EA be like in different eras and what would their priorities have been” would be an interesting thread.
1. ^
  I’m also not sure what all the AI researchers would do, but I guess it would involve reading and writing papers
2. ^
  Feels more in keeping with the thought experiment and the actual responsibilities of the modern movement to have EAs include people who are super impressed by papers they read by Jon Snow and Florence Nightingale and want to put that sort of analysis into practice in more places. Which still could have speeded up evidence based medicine a lot, but over half a century later...

David T 29 Nov 2024 15:54 UTC
20 points
4 ∶ 2
in reply to: Thomas Kwa’s comment on: Thomas Kwa’s Shortform
I suspect the net impact would be pretty low. Most of the really compelling consequentialist arguments like “if we don’t agree to this there will be a massive civil war in future” and “an Industrial Revolution will leave everyone far richer anyway” are future knowledge that your thought experiment strips people of. It didn’t take complex utility calculations to persuade people that slaves experienced welfare loss; it took deontological arguments versed [mainly] in religious belief to convince people that slaves were actually people whose needs deserved attention. And Jeremy Bentham was already there to offer utilitarian arguments, to the extent people were willing to listen to them.
And I suspect that whilst a poll of Oxford-educated utilitarian pragmatists with a futurist mindset transported back to 1776 would near-unanimously agree that slavery wasn’t a good thing, they’d probably devote far more of their time and money to stuff they saw as more tractable like infectious diseases and crop yields, writing some neat Benthamite literature and maybe a bit of wondering whether Newcomen engines and canals made the apocalypse more likely.
I can’t imagine the messy political compromise that was the US constitution being something that EAs would be skilled at negotiating even if they were in a position to do, so I don’t see why they’d be able to get rid of the Fugitive Slave Clause. They might have decided helping fugitive slaves was an effective way of helping people and concluded they had an ethical obligation to spend lots of money on it. Or concluded it was more cost-effective to leave them be and buy a few more slaves’ freedom legally instead...

David T 28 Nov 2024 12:26 UTC
1 point
1 ∶ 0
in reply to: Søren Elverlin’s comment on: What are some criticisms of PauseAI?
Limited tractability is easily outweighed by a very high potential impact.
This is the case only if you make the assumption that protesting is incapable of having a negative impact on outcomes either directly by creating a negative impact or indirectly by causing people supporting the other, richer and better connected side to put more effort into regulatory capture. Other people have more made specific claims about the nature of PauseAI’s campaigns; I’m pointing out that this is a battle where their expected outcome isn’t necessarily positive even if they’re pretty good...
(Relevant context: the incoming US administration is ambivalent at best towards the likes of Altman but is extremely hostile to doom and safety narratives to the point it sees partisan advantage in being seen to rejecting them in favour of economic growth; it also sees arms races as things to participate in and win.)
And even if one ignores the abundance of evidence that protest movements often have negative impacts (particularly in the short term, which is what PauseAI care about) and that this might be one of those cases, the Pascalian argument that the payoffs are so high the lack of tractability is irrelevant to its effectiveness only works if there’s literally no plausibly more effective ways to achieve the same goal.

David T 24 Nov 2024 11:45 UTC
14 points
4 ∶ 2
on: What are some criticisms of PauseAI?
What PauseAI wants to ban or “pause” seems fairly weakly defined and not necessarily relevant to any actual threat level. Their stated goals focus on banning scaling of LLM architecture with known limitations that make ‘takeover’ scenarios unlikely (limited context windows, lack of recursive self-updating independently from training, dependence on massive datacentres to run) and known problems (inscrutability and obvious lack of consistent “alignment”) that are still problems with smaller models if you try to use them for anything sensitive. It’s not clear what “more powerful than GPT4” actually means. Nor is it clear what the level of understanding that will result in un-pausing is or how it will be obtained without any models to study.
Banning LLMs of a certain scale might even have the perverse effect of encouraging companies to optimize performance or reinvent the idea of learning in other ways which are more risky. Or setting back ability to understand extremely powerful LLMs when someone develops them outside a US/EU legislative framework anyway. Or preventing positive AI developments that could save thousands of lives (or from the point of view of a longtermist that believes existential risk is currently nonzero including non-AI factors but might drop to zero in future because of friendly AI, perhaps 10^31 lives!)
Beyond that I think from the perspective of being an effective giving target, PauseAI suffers from the same shortcomings most lobbying outfits do (influencing government and public opinion in an opposing direction to economic growth is hard , it’s unclear what results a marginal dollar donation achieves, and the other side have a lot more dollars and connections to ramp up activity in an equal and opposite direction if they feel their business interests are threatened) so there’s no reason to believe they’re effective even if one agrees their goal is well-defined and correct.
You could also question the motivations of some of the people arguing for AI pauses (hi Elon, we see the LLM you launched shortly after signing the letter saying that LLMs that were ahead of yours were dangerous and should be banned...) although I don’t think this applies to the PauseAI organization specifically.