Quantumfreakonomics
No bio...
User ID: 324
Quantumfreakonomics
22hr ago
I still don't think this works out. The total addressable market for AI services in a regulated innovation-skeptical licensing regime is (to within an order of magnetude) $500 billion. The total addressable market in a world where Anthropic is free to pursue unregulated recursive self-improvement is ~all economic value in the universe.
Quantumfreakonomics
22hr ago
You know, this kind of makes sense for when Anthropic lobbies against open weights, or against Chinese model use, or in favor of chip export restrictions, but I don't really see how regulating frontier capabilities research helps Anthropic's valuation.
Quantumfreakonomics
23hr ago
AI X-Risk Goes Mainstream?
Jacob Coxon:
116,400,000 views, 634,000 likes.
Evan Hubinger (head of alignment at Anthropic):
34,000,000 views, 51,000 likes.
None of this is substantially different from what certain AI industry insiders have been saying for 5-ish years now, but it's clear from the replies, quote-tweets, and engagement statistics that most people are hearing about this for the first time. Yes, there was clearly some coordination with friendly media prior to the announcement given the almost simultaneous WSJ article (whereas news outlets were initially reluctant to publish anything on the unexpected news of OpenAI solving a Millennium Prize Problem). One can take the maximally conspiratorial position that this is a 4D bankshot by Democrats to end tech innovation forever, but it was going to break through eventually.
In other words, Coffeezilla is now on the case.
Quantumfreakonomics
1d ago
Apply the intermediate value theorem. If AI started 2026 much worse than human mathematicians, and (hypothetically) by the end of 2026 AI is much better than human mathematicians, then at some point in 2026 AI capabilities pass through the point at which they are useful but still need human mathematician assistance to reach their full potential.
I think that the draft proof was itself largely written by AI, and so the controversy is about which specific group of humans pressing the, “insert coins to solve mathematics”, button gets the credit.
EDIT: Also my understanding is that the draft proof was for a related-but-not-Millennium-Prize-worthy subproblem.
Quantumfreakonomics
2d ago
I got a news alert on my laptop after the result was announced. It was for the head of the Smithsonian resigning. There’s nothing on Fox News, nothing on CNN. The media is asleep at the wheel. Normies have no idea the man-made horrors beyond our comprehension that are being built as we speak.
Quantumfreakonomics
2d ago
I have no idea what OpenAI’s internal data pipelines look like, but I think it’s best to assume that anything you type into the chatbox can and will be used against you by Sam Altman/the government/The Basilisk.
Quantumfreakonomics
2d ago
Relevant paragraph:
”We (the researchers and the agents) did not see any of their work through any means until they released it publicly — in particular, no specific user data was accessed in order to solve this problem. While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models. However, our proofs differ significantly and even the precise results proved are different in the Euler case (forced vs unforced).”
Quantumfreakonomics
2d ago
The problem with the Nazis wasn't that they wanted to restrict immigration and expel foreign influence. It wasn't even (really) that they hated Jews. The problem with the Nazis was that they had an expansionist foreign policy.
What is the AfD position on German Irredentism? I expect that it's not in the official platform documents that they want to annex Austria and take back Königsberg, but what do the rank-and-file voters think of these questions?
Quantumfreakonomics
4d ago
Moral hazard. Sports team owners get the blame whenever the team is persistantly bad (or just persistantly mediocre), and unlike coaches or GMs, they can't be fired for poor performance. The only way to force-out an owner is to catch him in the midst of some impropriety. This makes sports team owners prime targets for TMZ-style exposes.
Quantumfreakonomics
5d ago
>tfw you can't commit suicide because you're still paralyzed from your previous suicide attempt.
Quantumfreakonomics
6d ago
I saw Matt Walsh compare this to the Karen Read case, which is an interesting angle, but it really doesn't make the point Matt thinks it does. I didn't watch the Clancy trial, but I did watch most of the Read trials. (For context, Karen Read was accused of hitting her boyfriend with her car after an argument while drunk) On a superficial culture-war angle, the cases are similar:
- They were both in Massachussets.
- Both Karen Read and Lindsay Clancy are middle-aged reasonably-photogenic white women.
- The first proceeding in both cases ended in a hung-jury mistrial.
- I would not be surprised if the people outside the Plymouth County Courthouse are the exact same people who were outside the Norfolk County Courthouse last year.
The reason this doesn't make Matt's point is that the Karen Read prosecution was a shitshow. The injuries to the victim's body didn't look like a car impact, all of the people who were at the location the body was found on the night of the incident were caught in a bunch of incredibly suspicious lies and omissions, and the investigators assigned to the case were incredibly unprofessional, biased, and plausibly planted evidence. The reputation of Massachussets law enfocement still hasn't recovered.
Quantumfreakonomics
6d ago
A speedrunning automaton picks Bulbasaur. A real AGI that understands truth, beauty, and The American Way would pick charmander.
Quantumfreakonomics
6d ago
I had trouble with Pokemon games when I was 6. Mt. Moon was an almost insurmountable challenge.
Quantumfreakonomics
6d ago
I want to know which starter the different models pick.
Quantumfreakonomics
6d ago
I don't know what any of these eval scores mean. I no longer have the ability to intuitively "feel" how much more intelligent each iterative model release is. I don't know how to code. I don't know how to do graduate-level math. These models have been smarter than me for quite some time.
Any researcher emotionally sane either:
Quit years ago because they understood the risks (Hinton, Kokotajlo, arguably Sutskever), or
Doesn't understand the potential of the technology well enough to make meaningful progress (LeCunn, Chollet).
Meta AI hires normal people. Look where that's gotten them.
More options
Context Copy link