David Hornbein

Karma: 1,173

David Hornbein 15 Apr 2024 17:55 UTC
27 points
4
in reply to: Ben Pace’s comment on: Anthropic AI made the right call
I’m coming to this late, but this seems weird. Do I understand correctly that many people were saying that Anthropic, the AI research company, had committed never to advance the state of the art of AI research, and they believed Anthropic would follow this commitment? That is just… really implausible.
This is the sort of commitment which very few individuals are psychologically capable of keeping, and which ~zero commercial organizations of more than three or four people are institutionally capable of keeping, assuming they actually do have the ability to advance the state of the art. I don’t know whether Anthropic leadership ever said they would do this, and if they said it then I don’t know whether they meant it earnestly. But even imagining they said it and meant it earnestly there is just no plausible world in which a company with hundreds of staff and billions of dollars of commercial investment would keep this commitment for very long. That is not the sort of thing you see from commercial research companies in hot fields.
If anyone here did believe that Anthropic would voluntarily refrain from advancing the state of the art in all cases, you might want to check if there are other things that people have told you about themselves, which you would really like to be true of them, but you have no evidence for other than their assertions, and would be very unusual if they were true.

David Hornbein 19 Mar 2024 18:26 UTC
1 point
17
in reply to: habryka’s comment on: Effective Aspersions: How the Nonlinear Investigation Went Wrong
Ben is working on a response, and given that I think it’s clearly the right call to wait a week or two until we have another round of counter-evidence before jumping to conclusions. If in a week or two people still think the section of “Avoidable, Unambiguous falsehoods” does indeed contain such things, then I think an analysis like this makes sense
This was three months ago. I have not seen the anticipated response. Setting aside the internal validity of your argument above, the promised counterevidence did not arrive in anything like a reasonable time.
TracingWoodgrains clearly made the right call in publishing, rather than waiting for you.

David Hornbein 19 Dec 2023 23:00 UTC
49 points
15
in reply to: Matt Goldenberg’s comment on: Effective Aspersions: How the Nonlinear Investigation Went Wrong
Yes, obviously, but they use different strategies. Male sociopaths rarely paint themselves as helpless victims because it is not an effective tactic for men. One does notice that, while the LW community is mostly male, ~every successful callout post against a LW community organization has been built on claims of harm to vulnerable female victims.

David Hornbein 19 Dec 2023 21:05 UTC
59 points
41
in reply to: habryka’s comment on: Effective Aspersions: How the Nonlinear Investigation Went Wrong
When you say “it’s clearly the right call to wait a week or two until we have another round of counter-evidence before jumping to conclusions”, is this a deliberate or accidental echo of the similar request from Nonlinear which you denied?
Like, on the deliberate way of reading this, the subtext is “While Lightcone did not wait a week or two for counter-evidence and still defends this decision, you should have waited in your case because that’s the standard you describe in your article.” Which would be a hell of a thing to say without explicitly acknowledging that you’re asking for different standards. (And would also misunderstand TracingWoodgrains’s actual standard, which is about the algorithm used and not how much clock time is elapsed, as described in their reply to your parent comment.) Or on the accidental way of reading this, the subtext is “I was oblivious to how being publicly accused of wrongdoing feels from the inside, and I request grace now that the shoe is on the other foot.” Either of these seems kind of incredible but I can’t easily think of another plausible way of reading this. I suppose your paragraph on wanting to take the time to make a comprehensive response (which I agree with) updates my guess towards “oblivious”.

David Hornbein 12 Dec 2023 23:49 UTC
22 points
4
on: Nonlinear’s Evidence: Debunking False and Misleading Claims
On Pace’s original post I wrote:
“think about how bad you expect the information would be if I selected for the worst, credible info I could share”
Alright. Knowing nothing about Nonlinear or about Ben, but based on the rationalist milieu, then for an org that’s weird but basically fine I’d expect to see stuff like ex-employees alleging a nebulously “abusive” environment based on their own legitimately bad experiences and painting a gestalt picture that suggests unpleasant practices but without any smoking-gun allegations of really egregious concrete behavior (as distinct from very bad effects on the accusers); allegations of nepotism based on social connections between the org’s leadership and their funders or staff; accusations of shoddy or motivated research which require hours to evaluate; sources staying anonymous for fear of “retaliation” but without being able to point to any legible instances of retaliation or concrete threats to justify this; and/or thirdhand reports of lying or misdirection around complicated social situations.
[reads post]
This sure has a lot more allegations of very specific and egregious behavior than that, yeah.
Having looked at the evidence and documentation which Nonlinear provides, it seems like the smoking-gun allegations of really egregious concrete behavior are probably just false. I have edited my earlier comment accordingly.

David Hornbein 12 Dec 2023 19:09 UTC
13 points
1
on: The likely first longevity drug is based on sketchy science. This is bad for science and bad for longevity.
This is a bit of a tangent, but is there a biological meaning to the term “longevity drug”? For a layman like me, my first guess is that it’d mean something like “A drug that mitigates the effects of aging and makes you live longer even if you don’t actively have a disease to treat.” But then I’d imagine that e.g. statins would be a “longevity drug” for middle-aged men with a strong family history of heart disease, in that it makes the relevant population less susceptible to an aging-related disease and thereby increases longevity, yet the posts talk about the prospect of creating the “first longevity drug” so clearly it’s being used in a way that doesn’t include statins. Is there a specific definition I’m ignorant of, or is it more of a loose marketing term for a particular subculture of researchers and funders, or what?

David Hornbein 11 Dec 2023 10:50 UTC
12 points
7
on: Principles For Product Liability (With Application To AI)
We can certainly debate whether liability ought to work this way. Personally I disagree, for reasons others have laid out here, but it’s fun to think through.
Still, it’s worth saying explicitly that as regards the motivating problem of AI governance, this is not currently how liability works. Any liability-based strategy for AI regulation must either work within the existing liability framework, or (much less practically) overhaul the liability framework as its first step.

David Hornbein 9 Dec 2023 23:09 UTC
5 points
0
in reply to: johnswentworth’s comment on: What I Would Do If I Were Working On AI Governance
Cars are net positive, and also cause lots of harm. Car companies are sometimes held liable for the harm caused by cars, e.g. if they fail to conform to legal safety standards or if they sell cars with defects. More frequently the liability falls on e.g. a negligent driver or is just ascribed to accident. The solution is not just “car companies should pay out for every harm that involves a car”, partly because the car companies also don’t capture all or even most of the benefits of cars, but mostly because that’s an absurd overreach which ignores people’s agency in using the products they purchase. Making cars (or ladders or knives or printing presses or...) “robust to misuse”, as you put it, is not the manufacturer’s job.
Liability for current AI systems could be a good idea, but it’d be much less sweeping than what you’re talking about here, and would depend a lot on setting safety standards which properly distinguish cases analogous to “Alice died when the car battery caught fire because of poor quality controls” from cases analogous to “Bob died when he got drunk and slammed into a tree at 70mph”.

David Hornbein 6 Dec 2023 4:25 UTC
3 points
0
on: How do you feel about LessWrong these days? [Open feedback thread]
It’s fun to come through and look for interesting threads to pull on. I skim past most stuff but there’s plenty of good and relevant writing to keep me coming back. Yeah sure it doesn’t do a super great job of living up to the grandiose ideals expressed in the Sequences but I don’t really mind, I don’t feel invested in ~the community~ that way so I’ll gladly take this site for what it is. This is a good discussion forum and I’m glad it’s here.

David Hornbein 28 Nov 2023 16:24 UTC
7 points
−1
in reply to: wassname’s comment on: OpenAI: Facts from a Weekend
Toner’s employer, the Center for Security and Emerging Technology (CSET), was founded by Jason Matheny. Matheny was previously the Director of the Intelligence Advanced Research Projects Activity (IARPA), and is currently CEO of the RAND Corporation. CSET is currently led by Dewey Murdick, who previously worked at the Department of Homeland Security and at IARPA. Much of CSET’s initial staff was former (or “former”) U.S. intelligence analysts, although IIRC they were from military intelligence rather than the CIA specifically. Today many of CSET’s researchers list prior experience with U.S. civilian intelligence, military intelligence, or defense intelligence contractors. Given the overlap in staff and mission, U.S. intelligence clearly and explicitly has a lot of influence at CSET, and it’s reasonable to suspect a stronger connection than that.
I don’t see it for McCauley though.

David Hornbein 27 Nov 2023 23:27 UTC
5 points
5
on: why did OpenAI employees sign
Suppose you’re an engineer at SpaceX. You’ve always loved rockets, and Elon Musk seems like the guy who’s getting them built. You go to work on Saturdays, you sometimes spend ten hours at the office, you watch the rockets take off and you watch the rockets land intact and that makes everything worth it.
Now imagine that Musk gets in trouble with the government. Let’s say the Securities and Exchange Commission charges him with fraud again, and this time they’re *really* going after him, not just letting him go with a slap on the wrist like the first time. SpaceX’s board of directors negotiates with SEC prosecutors. When they emerge they fire Musk from SpaceX, and remove Elon and Kimbal Musk from the board. They appoint Gwynne Shotwell as the new CEO.
You’re pretty worried! You like Shotwell, sure, but Musk’s charisma and his intangible magic have been very important to the company’s success so far. You’re not sure what will happen to the company without him. Will you still be making revolutionary new rockets in five years, or will the company regress to the mean like Boeing? You talk to some colleagues, and they’re afraid and angry. No one knows what’s happening. Alice says that the company would be nothing without Musk and rails at the board for betraying him. Bob says the government has been going after Musk on trumped-up charges for a while, and now they finally got him. Rumor has it that Musk is planning to start a new rocket company.
Then Shotwell resigns in protest. She signs an open letter calling for Musk’s reinstatement and the resignation of the board. Board member Luke Nosek signs it too, and says his earlier vote to fire Musk was a huge mistake.
You get a Slack message from Alice saying that she’s signed the letter because she has faith in Musk and wants to work at his company, whichever company that is, in order to make humanity a multiplanetary species. She asks if you want to sign.
How do you feel?

David Hornbein 22 Nov 2023 20:20 UTC
61 points
32
on: OpenAI: The Battle of the Board
I really don’t think you can justify putting this much trust in the NYT’s narrative of events and motivations here. Like, yes, Toner did publish the paper, and probably Altman did send her an email about it. Then the NYT article tacitly implies but *doesn’t explicitly say* this was the spark that set everything off, which is the sort of haha-it’s-not-technically-lying that I expect from the NYT. This post depends on that implication being true.

David Hornbein 21 Nov 2023 23:20 UTC
7 points
3
on: Am I going insane or is the quality of education at top universities shockingly low?
Yeah it’s pretty bad. There are some professors who are better than just reading the textbook, but unfortunately they’re the exceptions. My undergrad experience got a lot more productive once I started picking my courses based on the *teacher* more than on the *subject*.

David Hornbein 20 Nov 2023 17:55 UTC
13 points
3
in reply to: dr_s’s comment on: OpenAI: Facts from a Weekend
Maybe Shear was lying. Maybe the board lied to Shear, and he truthfully reported what they told him. Maybe “The board did *not* remove Sam over any specific disagreement on safety” but did remove him over a *general* disagreement which, in Sutskever’s view, affects safety. Maybe Sutskever wanted to remove Altman for a completely different reason which also can’t be achieved after a mass exodus. Maybe different board members had different motivations for removing Altman.

David Hornbein 20 Nov 2023 17:37 UTC
37 points
27
in reply to: dr_s’s comment on: OpenAI: Facts from a Weekend
My guess: Sutskever was surprised by the threatened mass exodus. Whatever he originally planned to achieve, he no longer thinks he can succeed. He now thinks that falling on his sword will salvage more of what he cares about than letting the exodus happen.

David Hornbein 5 Nov 2023 17:18 UTC
19 points
−4
on: Propaganda or Science: A Look at Open Source AI and Bioterrorism Risk
I note that the comments here include a lot of debate on the implications of this post’s thesis and on policy recommendations and on social explanations for why the thesis is true. No commenter has yet disagreed with the actual thesis itself, which is that this paper is a representative example of a field that is “more advocacy than science”, in which a large network of Open Philanthropy Project-funded advocates cite each other in a groundless web of footnotes which “vastly misrepresents the state of the evidence” in service of the party line.

David Hornbein 3 Nov 2023 18:38 UTC
19 points
4
on: If AGI is imminent, why can’t I hail a robotaxi?
I can hail a robotaxi. So can anyone living in San Francisco, Phoenix, Beijing, Shanghai, Guangzhou, Shenzhen, Chongqing or Wuhan. The barriers to wider rollout are political and regulatory, not technological.
Waymo cars, and I believe Apollo and Cruise as well, are “level 4” autonomous vehicles, i.e. there is no human involved in driving them whatsoever. There is a “human available to provide assistance” in roughly the same sense that a member of the American Automobile Association has an offsite human available to provide assistance in case of crashes, flat tires, etc.
I don’t see any reason to think AGI is imminent but this particular argument against it doesn’t go through. Robotaxi tech is very good and improving swiftly.

David Hornbein 15 Oct 2023 1:51 UTC
24 points
21
on: Announcing MIRI’s new CEO and leadership team
Seems also worth mentioning that Gretta Duleba and Eliezer Yudkowsky are in a romantic relationship, according to Yudkowsky’s Facebook profile.

David Hornbein 28 Sep 2023 20:39 UTC
3 points
0
in reply to: Ruby’s comment on: Petrov Day Retrospective, 2023 (re: the most important virtue of Petrov Day & unilaterally promoting it)
At the risk of butting in, I also didn’t participate because none of the options reflected my own views on the important virtue of Petrov Day, which is more like “Do not start a fucking nuclear war”. You can try to distill that down to nicely categorized principles and virtues, and some of those might well be good things on their own, but at this level of abstraction it doesn’t capture what’s special about Petrov Day to me.
Trying to file down the Petrov Day narrative into supporting some other hobbyhorse, even if it’s a wonderful hobbyhorse which I otherwise support like resisting social pressure, is a disservice to what Stanislav Petrov actually did. The world is richer and more complex than that.
I personally preferred the past Petrov Day events, with the button games and standoffs between different groups and all that. They didn’t perfectly reflect the exact dilemma Petrov faced, sure, but what could. They were messy and idiosyncratic and turned on weird specific details. That feels like a much closer reflection of what makes Petrov’s story compelling, to me. Maybe the later stages of this year’s event would’ve felt more like that, if I’d seen them at the time, but reading the description I suspect probably not.
I like that you guys are trying a bunch of different stuff and it’s fine if this one thing didn’t land for me.

David Hornbein 20 Sep 2023 19:54 UTC
1 point
0
in reply to: Linch’s comment on: Protest against Meta’s irreversible proliferation (Sept 29, San Francisco)
One very common pattern is, most people oppose a technology when it’s new and unfamiliar, then later once it’s been established for a little while and doesn’t seem so strange most people think it’s great.