It’s worth noting also that many academics who signed the statement may face adverse issues like reputational risk as well as funding cut to their research programs if AI safety becomes an official policy.
For a large number of them, these risks are worth far more than any possible gain from signing it.
When a large number of smart, reputable people, including many with expert knowledge and little or negative incentives to act dishonestly, put their names down like this, one should pay attention.
Added:
Paul Christiano, a brilliant theoretical CS researcher who switched to AI Alignment several years ago, put the risks of “doom” for humanity at 46%.
Subtract OpenAI, Google, StabilityAI and Anthropic affiliated researchers (who have a lot to gain) and not many academic signatories are left.
Notably missing representation from the Stanford NLP (edit: I missed that Diyi Yang is a signatory on first read) and NYU groups who’s perspective I’d also be interested in hearing.
Not committing one way or another regarding the intent with this but it’s not as diverse an academic crowd as the long list may suggest and for a lot of these names there are incentives to act dishonestly (not claiming that they are).
Even if it’s just Yoshua Bengio, Geoffrey Hinton, and Stuart Russell, we’d probably agree the risks are not negligible. There are quite a few researchers from Stanford, UC Berkeley, MIT, Carnegie Mellon, Oxford, Cambirdge, Imperial College, Edinburg, Tsinghua, etc who signed as well. Many of whom do not work for those companies.
We’re talking about nuclear war level risks here. Even a 1% chance should definitely be addressed. As noted above, Paul Christiano who has worked on AI risk and thought about it for a long time put it at 46%.
> There are quite a few researchers from Stanford, UC Berkeley, MIT, Carnegie Mellon, Oxford, Cambirdge, Imperial College, Edinburg, Tsinghua, etc who signed as well.
I know the Stanford researchers the most and the “biggest names” in LLMs from HAI and CRFM are absent. It would be useful to have their perspective as well.
I’d throw MetaAI in the mix as well.
Merely pointing out that healthy skepticism here is not entirely unwarranted.
> We’re talking about nuclear war level risks here.
LLM is already a misnomer. Latest versions are multimodal. Current versions can be used to build agents with limited autonomy. Future versions of LLMs are most likely capable of more independence.
Even dumb viruses have caused catastrophic harm. Why? It’s capable of rapid self replication in a massive number of existing vessels. You add in some intelligence, vast store of knowledge, huge bandwidth, and some aid by malicious human actors, what could such a group of future autonomous agents do?
I mean a small group of malicious humans can already bioengineer a deadly virus with CRISPR and open source tech without AI.
This is hardly the first time in history a new technological advancement may be used for nefarious purposes.
It’s a discussion worth having as AI advances but if [insert evil actor] wants to cause harm there are many cheaper and easier ways to do this right now.
To come out and say we need government regulation today does stink at least a little bit of protectionism as practically speaking the “most evil actors” would not adhere to whatever is being proposed, but this would impact the competitive landscape and the corporations yelling the loudest right now have the most to gain, perhaps coincidence but worth questioning.
> I mean a small group of malicious humans can already bioengineer a deadly virus with CRISPR and open source tech without AI.
That's what interesting to me. People fearmongering about bioengineering and GMO's were generally dismissed as being anti-science and holding humankind back (or worse, that there opposition to progress meant they had blood on their hands). Yet many of the people who mocked them proved themselves to be even more dogmatic and apocalyptic, while being much closer to influencing regulations. And the technology they're fear-mongering about is even further from being able to harm people than biotech is. We are actually able to create harmful biotech today if we want; we don't know when we'll ever be able to create AGI, and if it would even pose a danger if we did.
This mentality - "there could be a slight chance research into this could eventually lead to apocalyptic technology, no I don't have any idea how but the danger is so great we need a lot of regulation" - would severely harm scientific growth if we applied it consistently. Of course everyone is going to say "the technology I'm afraid of is _actually_ dangerous, the technology they're afraid if isn't." But we honestly have no clue when we're talking about technology that we have no idea how to create at the moment.
Counterpoint: CRISPR only reignited what was already a real fear of reduced difficulty and costs of engineering deadly pathogens.
In fact, what you and GP wrote is baffling to me. The way I see it, biotech is stupidly obviously self-evidently dangerous, because let's look at the facts:
- Genetic engineering gets easier and cheaper and more "democratized"; in the last 10 years, the basics were already accessible to motivated schools and individual hobbyists;
- We already know enough, with knowledge accessible at the hobbyist level, to know how to mix and match stuff and get creative - see "synthetic biology";
- The substrate we're working with is self-replicating molecular nanotechnology; more than that, it's usually exactly the type that makes people get sick - bacteria (because they're most versatile nanobots) and viruses (because they're natural code injection systems).
Above is the "inside view"; for "outside view", I'll just say this: the fact that "lab leak" hypothesis of COVID-19 was (or still is?) considered to be one of the most likely explanations for the pandemics already tells you that the threat is real, and consequences are dire.
I don't know how can you possibly look at that and conclude "nah, not dangerous, needs to be democratized so the Evil Elites don't hoard it all".
There must be some kind of inverse "just world fallacy" fallacy of blaming everything on evil elites and 1%-ers that are Out To Get Us. Or maybe it's just another flavor of the NWO conspiracy thinking, except instead the Bildenbergs and the Jews its Musk, Bezos and the tech companies.
Same is, IMHO, with AI. Except that one is more dangerous because it's a technology-using technology - that is, where e.g. accidentally or intentionally engineered pathogens could destroy civilization directly, AI could do it by using engineered pathogens - or nukes, or mass manipulation, or targeted manipulation, or ... countless other things.
EDIT:
And if you ask "why, if it's really so easy to access and dangerous, we haven't already been killed by engineered pathogens?", the answer is a combination of:
1. vast majority of people not bearing ill intent;
2. vast majority of people being not interested and not able to perform (yet!) this kind of "nerdy thing";
3. a lot of policing and regulatory attention given to laboratories and companies playing with anything that could self-replicate and spread rapidly;
4. well-developed policies and capacity for dealing with bio threats (read: infectious diseases, and diseases in general);
5. this being still new enough that the dangerous and the careless don't have an easy way to do what in theory they already could.
Note that despite 4. (and 3., if you consider "lab leak" a likely possibility), COVID-19 almost brought the world down.
Great points. Will just add a point 1.5: There's usually an inverse correlation between ill intent and competence, so the subset of people who both want to cause harm to others on a mass scale and who are also able to pull it off is small
I’m not sure there is a way for someone to engineer a deadly virus while completely innoculating themselves from it.
Short-term AI risk likely comes from a mix of malicious intent and further autonomy that causes harm the perpetrators did not expect. In the longer run, there is a good chance of real autonomy and completely unexpected behaviors from AI.
Why do you have to inoculate yourself from it to create havoc? Your analogy of “nuclear war” also has no vaccine.
AI autonomy is a hypothetical existential risk, especially in the short term. There are many non-hypothetical existential risks including actual nuclear proliferation and escalating great power conflicts happening right now.
Again my point being that this is an important discussion but appears overly dramatized, just like there are people screaming doomsday there are also equally qualified people (like Yann LeCun) screaming BS.
But let’s entertain this for a second, can you posit a hypothetical where in the short term a nefarious actor can abuse AI or autonomy results in harm? How does this compare to non-AI alternatives for causing harm?
This gets countered by running one (or more) of those same amazing autonomous agents locally for your own defense. Everyone's machine is about to get much more intelligent.
I used LLM because the people shouting the loudest come from a LLM company which claimed their newest language model can be used to create bioweapons in their whitepaper.
Semantics aside the recent interest in AI risk was clearly stimulated by LLMs and the camp that believes this is the path to AGI which may or may not be true depending who you ask.
I can only imagine Eleizer Yudkowsky and Rob Miles looking on this conversation with a depressed scream and a facepalm respectively.
They've both been loudly concerned about optimisers doing over-optimisation, and society having a Nash equilibrium where everyone's using them as hard as possible regardless of errors, since before it was cool.
To the extent that this may be true (I've not exactly studied which public thinkpiece writers care about which AI so it might easily be the image generators that get this crown for all I know), that's because ChatGPT actually does something that a normal person understands.
A paper titled "Dual use of artificial-intelligence-powered drug discovery" (published last year) got a few angst pieces and is mostly forgotten by the general public and media so far as I can tell; but the people behind it both talked directly to regulators and other labs to help advise them how many other precursor chemicals were now potential threats, and also went onto the usual podcasts and other public forums to raise awareness of the risk to other AI researchers.
The story behind that was "ugh, they want us to think about risks… what if we ask it to find dangerous chemicals instead of safe ones? *overnight* oh no!"
> I can only imagine Eleizer Yudkowsky and Rob Miles looking on this conversation with a depressed scream and a facepalm respectively.
Whenever Yudkowsky comes up on my Twitter feed I'm left with an impression that I'm not going to have any more luck conversing AI with those in his orbit than I am discussing the rapture with a fundamentalist Christian. For example, the following Tweet[1]. If a person believes this is from a deep thinker that should be taken very seriously rather than an unhinged nutcase, our worldviews are probably too far apart to ever reach a common understanding:
> Fools often misrepresent me as saying that superintelligence can do anything because magic. To clearly show this false, here's a concrete list of stuff I expect superintelligence can or can't do:
> - FTL (faster than light) travel: DEFINITE NO
> - Find some hack for going >50 OOM past the amount of computation that naive calculations of available negentropy would suggest is possible within our local volume: PROBABLE NO
> - Validly prove in first-order arithmetic that 1 + 1 = 5: DEFINITE NO
> - Prove a contradiction from Zermelo-Frankel set theory: PROBABLE NO
> - Using current human technology, synthesize a normal virus (meaning it has to reproduce itself inside human cells and is built of conventional bio materials) that infects over 50% of the world population within a month: YES
> (note, this is not meant as an argument, this is meant as a concrete counterexample to people who claim 'lol doomers think AI can do anything just because its smart' showing that I rather have some particular model of what I roughly wildly guess to be a superintelligence's capability level)
> - Using current human technology, synthesize a normal virus that infects 90% of Earth within an hour: NO
> - Write a secure operating system on the first try, zero errors, no debugging phase, assuming away Meltdown-style hardware vulnerabilities in the chips: DEFINITE YES
> - Write a secure operating system for actual modern hardware, on the first pass: YES
> - Train an AI system with capability at least equivalent to GPT-4, from the same dataset GPT-4 used, starting from at most 50K of Python code, using 1000x less compute than was used to train GPT-4: YES
> - Starting from current human tech, bootstrap to nanotechnology in a week: YES
> - Starting from current human tech, bootstrap to nanotechnology in an hour: GOSH WOW IDK, I DON'T ACTUALLY KNOW HOW, BUT DON'T WANT TO CLAIM I CAN SEE ALL PATHWAYS, THIS ONE IS REALLY HARD FOR ME TO CALL, BRAIN LEGIT DOESN'T FEEL GOOD BETTING EITHER WAY, CALL IT 50:50??
> - Starting from current human tech and from the inside of a computer, bootstrap to nanotechnology in a minute: PROBABLE NO, EVEN IF A MINUTE IS LIKE 20 SUBJECTIVE YEARS TO THE SI
> - Bootstrap to nanotechnology via a clean called shot: all the molecular interactions go as predicted the first time, no error-correction rounds needed: PROBABLY YES but please note this is not any kind of necessary assumption because It could just build Its own fucking lab, get back the observations, and do a debugging round; and none of the processes there intrinsically need to run at the speed of humans taking hourly bathroom breaks, it can happen at the speed of protein chemistry and electronics. Please consider asking for 6 seconds how a superintelligence might possibly overcome such incredible obstacles of 'I think you need a positive nonzero number of observations', for example, by doing a few observations, and then further asking yourself if those observations absolutely have to be slow like a sloth
> - Bootstrap to nanotechnology by any means including a non-called shot where the SI designs more possible proteins than It needs to handle some of the less certain cases, and gets back some preliminary observations about how they interacted in a liquid medium, before it actually puts together the wetware lab on round 2: YES
(The Tweet goes on, you can read the rest of it at the link below, but that should give you the gist.)
I don't have twitter and I agree his tweets have an aura of lunacy, which is a shame as he's quite a lot better as a long-form writer. (Though I will assume his long-form writings about quantum mechanics is as bad as everyone else unless a physicist vouches for them).
But, despite that, I don't understand why you chose that specific example — how is giving a list of what he thinks an AI probably can and can't do, in the context of trying to reduce risks because he thinks loosing is the default, similar to a fundamentalist Christian who wants to immanentize the eschaton because the idea the good guys might lose when God is on their side is genuinely beyond comprehension?
Some of the academics who signed are either not doing AI research e.g climatologists, genomics, philosophy. Or they have Google connections that aren't disclosed. E.g. Peter Norvig is listed as Stanford University but ran Google Research for many years, McIlrath is associated with the Vector Institute which is funded by Google.
I just took that list and separated everyone that had any commercial tie listed, regardless of the company. 35 did and 63 did not.
> "Subtract OpenAI, Google, StabilityAI and Anthropic affiliated researchers (who have a lot to gain) and not many academic signatories are left."
You're putting a lot of effort into painting this list in a bad light without any specific criticism or evidence of malfeasance. Frankly, it sounds like FUD to me.
With corporate conflicts (that I recognized the names of):
Yoshua Bengio: Professor of Computer Science, U. Montreal / Mila,
Victoria Krakovna: Research Scientist, Google DeepMind,
Mary Phuong: Research Scientist, Google DeepMind,
Daniela Amodei: President, Anthropic,
Samuel R. Bowman: Associate Professor of Computer Science, NYU and Anthropic,
Helen King: Senior Director of Responsibility & Strategic Advisor to Research, Google DeepMind,
Mustafa Suleyman: CEO, Inflection AI,
Emad Mostaque: CEO, Stability AI,
Ian Goodfellow: Principal Scientist, Google DeepMind,
Kevin Scott: CTO, Microsoft,
Eric Horvitz: Chief Scientific Officer, Microsoft,
Mira Murati: CTO, OpenAI,
James Manyika: SVP, Research, Technology & Society, Google-Alphabet,
Demis Hassabis: CEO, Google DeepMind,
Ilya Sutskever: Co-Founder and Chief Scientist, OpenAI,
Sam Altman: CEO, OpenAI,
Dario Amodei: CEO, Anthropic,
Shane Legg: Chief AGI Scientist and Co-Founder, Google DeepMind,
John Schulman: Co-Founder, OpenAI,
Jaan Tallinn: Co-Founder of Skype,
Adam D'Angelo: CEO, Quora, and board member, OpenAI,
Simon Last: Cofounder & CTO, Notion,
Dustin Moskovitz: Co-founder & CEO, Asana,
Miles Brundage: Head of Policy Research, OpenAI,
Allan Dafoe: AGI Strategy and Governance Team Lead, Google DeepMind,
Jade Leung: Governance Lead, OpenAI,
Jared Kaplan: Co-Founder, Anthropic,
Chris Olah: Co-Founder, Anthropic,
Ryota Kanai: CEO, Araya, Inc.,
Clare Lyle: Research Scientist, Google DeepMind,
Marc Warner: CEO, Faculty,
Noah Fiedel: Director, Research & Engineering, Google DeepMind,
David Silver: Professor of Computer Science, Google DeepMind and UCL,
Lila Ibrahim: COO, Google DeepMind,
Marian Rogers Croak: VP Center for Responsible AI and Human Centered Technology, Google
Without:
Geoffrey Hinton: Emeritus Professor of Computer Science, University of Toronto,
Dawn Song: Professor of Computer Science, UC Berkeley,
Ya-Qin Zhang: Professor and Dean, AIR, Tsinghua University,
Martin Hellman: Professor Emeritus of Electrical Engineering, Stanford,
Yi Zeng: Professor and Director of Brain-inspired Cognitive AI Lab, Institute of Automation, Chinese Academy of Sciences,
Xianyuan Zhan: Assistant Professor, Tsinghua University,
Anca Dragan: Associate Professor of Computer Science, UC Berkeley,
Bill McKibben: Schumann Distinguished Scholar, Middlebury College,
Alan Robock: Distinguished Professor of Climate Science, Rutgers University,
Angela Kane: Vice President, International Institute for Peace, Vienna; former UN High Representative for Disarmament Affairs,
Audrey Tang: Minister of Digital Affairs and Chair of National Institute of Cyber Security,
Stuart Russell: Professor of Computer Science, UC Berkeley,
Andrew Barto: Professor Emeritus, University of Massachusetts,
Jaime Fernández Fisac: Assistant Professor of Electrical and Computer Engineering, Princeton University,
Diyi Yang: Assistant Professor, Stanford University,
Gillian Hadfield: Professor, CIFAR AI Chair, University of Toronto, Vector Institute for AI,
Laurence Tribe: University Professor Emeritus, Harvard University,
Pattie Maes: Professor, Massachusetts Institute of Technology - Media Lab,
Peter Norvig: Education Fellow, Stanford University,
Atoosa Kasirzadeh: Assistant Professor, University of Edinburgh, Alan Turing Institute,
Erik Brynjolfsson: Professor and Senior Fellow, Stanford Institute for Human-Centered AI,
Kersti Kaljulaid: Former President of the Republic of Estonia,
David Haussler: Professor and Director of the Genomics Institute, UC Santa Cruz,
Stephen Luby: Professor of Medicine (Infectious Diseases), Stanford University,
Ju Li: Professor of Nuclear Science and Engineering and Professor of Materials Science and Engineering, Massachusetts Institute of Technology,
David Chalmers: Professor of Philosophy, New York University,
Daniel Dennett: Emeritus Professor of Philosophy, Tufts University,
Peter Railton: Professor of Philosophy at University of Michigan, Ann Arbor,
Sheila McIlraith: Professor of Computer Science, University of Toronto,
Lex Fridman: Research Scientist, MIT,
Sharon Li: Assistant Professor of Computer Science, University of Wisconsin Madison,
Phillip Isola: Associate Professor of Electrical Engineering and Computer Science, MIT,
David Krueger: Assistant Professor of Computer Science, University of Cambridge,
Jacob Steinhardt: Assistant Professor of Computer Science, UC Berkeley,
Martin Rees: Professor of Physics, Cambridge University,
He He: Assistant Professor of Computer Science and Data Science, New York University,
David McAllester: Professor of Computer Science, TTIC,
Vincent Conitzer: Professor of Computer Science, Carnegie Mellon University and University of Oxford,
Bart Selman: Professor of Computer Science, Cornell University,
Michael Wellman: Professor and Chair of Computer Science & Engineering, University of Michigan,
Jinwoo Shin: KAIST Endowed Chair Professor, Korea Advanced Institute of Science and Technology,
Dae-Shik Kim: Professor of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST),
Frank Hutter: Professor of Machine Learning, Head of ELLIS Unit, University of Freiburg,
Scott Aaronson: Schlumberger Chair of Computer Science, University of Texas at Austin,
Max Tegmark: Professor, MIT, Center for AI and Fundamental Interactions,
Bruce Schneier: Lecturer, Harvard Kennedy School,
Martha Minow: Professor, Harvard Law School,
Gabriella Blum: Professor of Human Rights and Humanitarian Law, Harvard Law,
Kevin Esvelt: Associate Professor of Biology, MIT,
Edward Wittenstein: Executive Director, International Security Studies, Yale Jackson School of Global Affairs, Yale University,
Karina Vold: Assistant Professor, University of Toronto,
Victor Veitch: Assistant Professor of Data Science and Statistics, University of Chicago,
Dylan Hadfield-Menell: Assistant Professor of Computer Science, MIT,
Mengye Ren: Assistant Professor of Computer Science, New York University,
Shiri Dori-Hacohen: Assistant Professor of Computer Science, University of Connecticut,
Jess Whittlestone: Head of AI Policy, Centre for Long-Term Resilience,
Sarah Kreps: John L. Wetherill Professor and Director of the Tech Policy Institute, Cornell University,
Andrew Revkin: Director, Initiative on Communication & Sustainability, Columbia University - Climate School,
Carl Robichaud: Program Officer (Nuclear Weapons), Longview Philanthropy,
Leonid Chindelevitch: Lecturer in Infectious Disease Epidemiology, Imperial College London,
Nicholas Dirks: President, The New York Academy of Sciences,
Tim G. J. Rudner: Assistant Professor and Faculty Fellow, New York University,
Jakob Foerster: Associate Professor of Engineering Science, University of Oxford,
Michael Osborne: Professor of Machine Learning, University of Oxford,
Marina Jirotka: Professor of Human Centred Computing, University of Oxford
So the most “notable” AI scientists on this list have clear corporate conflicts. Some are more subtle:
> Geoffrey Hinton: Emeritus Professor of Computer Science, University of Toronto,
He’s affiliated with Vector (as well as some of the other Canadians on this list) and was at Google until very recently (unsure if he retained equity which would require disclosure in academia).
Hence my interest in disclosures as the conflicts are not always obvious.
How is saying that they should have disclosed a conflict that they did not disclose not accusatory? If that's the case, the accusation is entirely justified and should be surfaced! The other signatories would certainly want to know if they were signing in good faith when others weren't. This is what I need interns for.
I never said “they should have disclosed a conflict they did not disclose.”
Disclosures are absent from this initiative, some signatories have self-identified their affiliation by their own volition and even for those it is not in the context of a conflict disclosure.
There is no “signatories have no relevant disclosures” statement for those who did not for the omission to be malfeasance and pointing out the absence of a disclosure statement is not accusatory of the individuals, rather that the initiative is not transparent about potential conflicts.
Once again, it is standard practice in academia to make a disclosure statement if lecturing or publishing. While it is not mandatory for initiatives calling for regulation it would be nice to have.
I'd guess that a given academic isn't going to face much of a career risk for signing a statement also signed by other very prestigious academics, just the opposite. There's no part of very divided US political spectrum that I can see denouncing AI naysayers, unlike the scientists who signed anti-nuclear statements in 1960s or even people warning about global warming now (indeed, I'd guess the statement doesn't mention climate change 'cause it's still a sore point).
Moreover, talking about existential risk involves the assumption the current tech is going to continue to affect more and more fields rather than peaking at some point - this assumption guarantees more funding along with funding for risk.
All that said, I don't necessarily think the scientists involved are insincere. Rather, I would expect they're worried and signed this vague statement because it was something that might get traction. While the companies indeed may be "genuine" in the sense they're vaguely [concerned - edit] and also self-serving - "here's a hard problem it's important to have us wise, smart people in charge of and profiting from"
In interviews, Geoffrey Hinton and Yoshua Bengio certainly expressed serious concerns and even some plausible regret to their life’s work. They did not say anything that can be interpreted as your last sentence suggests at all.
My last sentence currently: "While the companies indeed may be "genuine" in the sense they're vaguely and also self-serving - "here's a hard problem it's important to have us wise, smart people in charge of and profiting from" - IE, I am not referring to the academics there.
I'm going to edit the sentence to fill in some missing words but I don't think this will change the meaning involved.
On the contrary, I suspect "How do we prevent our AIs from killing everyone?" will be a major research question with a great deal of funding involved. Plus, no one seems to be suggesting things like the medical ethics field or institutional review boards, which might have deleterious impacts on their work.
46% is such a precise number. This is why I can't take "rationalists", the Yudkowskys, and the Silvers seriously. Colossal assumptions turned into confidently stated probabilities.
You're putting a weirdly large amount of trust into, functionally, some dude who posted on lesswrong. Sure he has a PhD and is smart, but so is basically everyone else in the field, not just in alignment, and the median person in the field thinks the risk of "doom" is 2-5% (and that's conditioned on the supposed existence of a high level machine intelligence that the median expert believes might exist in 40 years). That still might be higher than you'd like, but it's not actually a huge worry in the grand scheme of things.
Like, if I told you that in 40 years, there was a 50% chance of something existing that had a 2% chance of causing extreme harm to the human population, I'm actually not sure that thing should be the biggest priority. Other issues may have more than a 1% chance of leading to terrible outcomes sooner.
The median person in field thinks 5-10%, not 2-5%, and median timelines are shorter than 40 years.
But this is all a distraction, since unaligned ASI is the default case absent significant efforts (that we aren't currently making), and trying to evaluate risk by averaging out the views of people who haven't actually explored the arguments very carefully (vs. just evaluating the arguments yourself) is a doomed effort.
> The median person in field thinks 5-10%, not 2-5%
The median person in the study here, under a particular definition was 5-10%, other comparable studies have found 2%, and similar questions using arguably better definitions in the same study found lower percentages.
> median timelines are shorter than 40 years.
The median person suggested a 50% chance in 39 years.
> since unaligned ASI is the default case
I challenge this assertion. Many relatively smart scholars who are more involved in the alignment space than, presumably either you or I, have put forth cogent arguments that alignment-by-default is perfectly reasonable. Dismissing those out of hand seems naive.
I work in the space (doing tech stuff that isn't direct research). The best argument I've seen for alignment by default is something like "morality comes from training data, and therefore the fact that LLM training data sets contain human moral intuitions will mean that an ASI stemming from such a training regime will share enough human values" (Quintin Pope believes something like this, as far as I can tell), which is deeply unconvincing, since it contradicts the evidence we _do_ have from human & animal value formation.
Happy to entertain other arguments that alignment-by-default is reasonable; most arguments I've seen are much worse than that one. I haven't seen many people make an active case for alignment-by-default, so much as leave open a whole bunch of swath of uncertainty for unknown unknowns.
For a large number of them, these risks are worth far more than any possible gain from signing it.
When a large number of smart, reputable people, including many with expert knowledge and little or negative incentives to act dishonestly, put their names down like this, one should pay attention.
Added:
Paul Christiano, a brilliant theoretical CS researcher who switched to AI Alignment several years ago, put the risks of “doom” for humanity at 46%.
https://www.lesswrong.com/posts/xWMqsvHapP3nwdSW8/my-views-o...