Joseph Miller

• Joseph Miller 15 Jan 2026 16:47 UTC

  4 points

  2

  in reply to: Joseph Miller’s comment on: Joseph Miller’s Shortform

  Or even better, at the the time when a post is frontpaged, check if will actually appear on the frontpage. If it is too old and has too little karma to be seen, then use the time-since-frontpaged.

• Joseph Miller 15 Jan 2026 14:20 UTC

  2 points

  0

  in reply to: Lucius Bushnaq’s comment on: Sim­ple sum­mary of AI Safety laws

  US laws:

  Well firstly note that the RAISE act has $1M - $3M escalating fines explicitly built in.

  But yes, my understanding is that if you pay up, you can carry on. However, they can be fined “per violation” and the law does not precisely define what this means. So the Attorney General can argue that each missed requirement /​ each non-complying model is a separate violation. Or if they’re feeling very aggressive, the AG could try to say that each day of non-compliance is a separate violation, but that may not hold up in court.

  This is where injunctive relief could come in, to directly order a company to retract a model. As noted, this was only explicitly included in the RAISE act^[1], and that looks set to be removed. However, AGs can often get injunctions even without explicit statutory language.

  EU AI Act:

  No, the consequences will escalate. Fines can be periodic, the maximum fine is huge and the length of the infraction is a factor in determining the penalty.

  And on top of that, companies have a duty to take “immediate corrective actions”. The Commission (with advice from the AI Office) can issue administrative orders requiring the withdrawal of a model. And ignoring that can get you into very heavy fines or criminal penalties.

  1. ^

     Except on whistleblower protections, where SB 53 explicitly includes injunctive relief.

• Joseph Miller 15 Jan 2026 3:15 UTC

  6 points

  2

  in reply to: habryka’s comment on: Joseph Miller’s Shortform

  Perhaps it could use time-since-frontpaged only if the karma is below some threshold.

• Joseph Miller 14 Jan 2026 23:42 UTC

  29 points

  18

  on: Joseph Miller’s Shortform

  On LessWrong, the frontpage algorithm down-weights older posts based on the time-since-posted, not the time-since-frontpaged. So, if a post doesn’t get frontpaged until a few days after posting, then it’s unlikely to get many views.

  LessWrong has an autofrontpager that works a reasonable amount of the time. Otherwise, posts have to be manually frontpaged by a person. In my experience, this was always quite quick, but my most recent post was not frontpaged until 3 days after it was posted, so AFAICT it never actually appeared on the frontpage (unless you clicked “Load More”).

  I think the solution is to downweight posts based on the time-since-frontpaged.

• Joseph Miller 7 Jan 2026 22:03 UTC

  2 points

  0

  in reply to: DanielFilan’s comment on: Sim­ple sum­mary of AI Safety laws

  Strictly speaking, the version that passed the senate has now been put into law.

  But to answer your question, no, this post describes what the bill will be like after the agreed upon “chapter amendments” have been implemented.

  Footnote 4 mentions one of the changes that is expected to be made.

Sim­ple sum­mary of AI Safety laws

• Joseph Miller 4 Jan 2026 23:52 UTC

  8 points

  5

  on: How I stopped be­ing sure LLMs are just mak­ing up their in­ter­nal ex­pe­rience (but the topic is still con­fus­ing)

  I would prefer this post if it didn’t talk about consciousness /​ internal experience. The issue is whether the LLM has some internal algorithms that are somehow similar /​ isomorphic to those in human brains.
  
  As it stands this posts implicitly assumes the Camp 2^[1] view of consciousness which I and many others find to be deeply sus. But the arguments the post puts forward are still relevant from a Camp 1 point of view, they just answer a question about algorithms, not about qualia.

  1. ^

     Quoting the key section of the linked post:

     Camp #1 tends to think of consciousness as a non-special high-level phenomenon. Solving consciousness is then tantamount to solving the Meta-Problem of consciousness, which is to explain why we think/​claim to have consciousness. In other words, once we’ve explained the full causal chain that ends with people uttering the sounds kon-shush-nuhs, we’ve explained all the hard observable facts, and the idea that there’s anything else seems dangerously speculative/​unscientific. No complicated metaphysics is required for this approach.

     Conversely, Camp #2 is convinced that there is an experience thing that exists in a fundamental way. There’s no agreement on what this thing is – some postulate causally active non-material stuff, whereas others agree with Camp #1 that there’s nothing operating outside the laws of physics – but they all agree that there is something that needs explaining. Therefore, even if consciousness is compatible with the laws of physics, it still poses a conceptual mystery relative to our current understanding. A complete solution (if it is even possible) may also have a nontrivial metaphysical component.

• Joseph Miller 9 Nov 2025 23:12 UTC

  4 points

  2

  in reply to: Vladimir_Nesov’s comment on: Condensation

  Even if all value is in computations (eg. everyone lives inside simulations), wouldn’t you have the same problems, just one level down? The physical world may be a type of computation and the computations may resemble the physical world.

• Joseph Miller 29 Oct 2025 15:40 UTC

  2 points

  0

  on: Strat­ified Utopia

  So if you wanted to live forever (level 2) or have a mechanical mind /​ body (level 5), would you have to leave Earth?

• Joseph Miller 17 Oct 2025 15:20 UTC

  5 points

  0

  in reply to: MichaelDickens’s comment on: Mikhail Samin’s Shortform

  Datapoint: I spoke to one Horizon fellow a couple of years ago and they did not care about x-risk.

• Joseph Miller 26 Sep 2025 16:44 UTC

  3 points

  0

  in reply to: snav’s comment on: snav’s Shortform

  Nice! I would like to see a visual showing the full decision tree. I think that would be even better for clarifying the different views of consciousness.

• Joseph Miller 26 Sep 2025 13:34 UTC

  7 points

  0

  in reply to: samuelshadrach’s comment on: xpostah’s Shortform

  Also mentioned in The Verge. (unpaywalled)

• Joseph Miller 12 Sep 2025 0:59 UTC

  21 points

  10

  in reply to: lc’s comment on: lc’s Shortform

  It doesn’t matter what IQ they have or how rational they were in 2005

  This is a reference to Eliezer, right? I really don’t understand why he’s on Twitter so much. I find it quite sad to see one of my heroes slipping into the ragebait Twitter attractor.

• Joseph Miller 30 Aug 2025 0:54 UTC

  5 points

  0

  in reply to: Zach Stein-Perlman’s comment on: Zach Stein-Per­l­man’s Shortform

  To clarify, the primary complaint from my perspective is not that they published the report a month after external deployment per se, but that the timing of the report indicates that they did not perform thorough pre-deployment testing (and zero external testing).

  And the focus on pre-deployment testing is not really due to any opinion about the relative benefits of pre- vs. post- deployment testing, but because they committed to doing pre-deployment testing, so it’s important that they in fact do pre-deployment testing.

• Joseph Miller 30 Aug 2025 0:39 UTC

  13 points

  5

  in reply to: Zach Stein-Perlman’s comment on: Zach Stein-Per­l­man’s Shortform

  3. Other companies are doing far worse in this dimension. At worst Google is 3rd-best in publishing eval results. Meta and xAI are far worse.

  Some reasons for focusing Google DeepMind in particular:

  • The letter was organized by PauseAI UK and signed by UK politicians. GDM is the only frontier AI company headquartered in the UK.

  • Meta and xAI already have a bad reputation for their safety practices, while GDM had a comparatively good reputation and most people were unaware of their violation of the Frontier AI Safety Commitments.

• Joseph Miller 30 Aug 2025 0:32 UTC

  3 points

  0

  in reply to: Zach Stein-Perlman’s comment on: Zach Stein-Per­l­man’s Shortform

  1. Sharing information on capabilities is good but public deployment is a bad time for that, in part because most risk comes from internal deployment.

  I’m not sure why this would make you not feel good about the critique or implicit ask of the letter. Sure, maybe internal deployment transparency would be better, but public deployment transparency is better than nothing.

  And that’s where the leverage is right now. Google made a commitment to transparency about external deployments, not internal deployments. And they should be held to that commitment or else we establish the precedent that AI safety commitments don’t matter and can be ignored.

• Joseph Miller 30 Aug 2025 0:31 UTC

  26 points

  −2

  in reply to: Zach Stein-Perlman’s comment on: Zach Stein-Per­l­man’s Shortform

  2. Google didn’t necessarily even break a commitment? The commitment mentioned in the article is to “publicly report model or system capabilities.” That doesn’t say it has to be done at the time of public deployment.

  This document linked on the open letter page gives a precise breakdown of exactly what the commitments were and how Google broke them (both in spirit and by the letter).^[1] The summary is this:

  • Google violated the spirit of commitment I by publishing its first safety report almost a
    month after public availability and not mentioning external testing in their initial report.

  • Google explicitly violated commitment VIII by not stating whether governments
    are involved in safety testing, even after being asked directly by reporters.

  But in fact the letter actually understates the degree to which Google DeepMind violated the commitments. The real story from this article is that GDM confirmed to Time that they didn’t provide any pre-deployment access to UK AISI:

  However, Google says it only shared the model with the U.K. AI Security Institute after Gemini 2.5 Pro was released on March 25.

  If UK AISI doesn’t have pre-deployment access, a large portion of their whole raison d’être is nullified.

  Google withholding access is quite strongly violating the spirit of commitment I of the Frontier AI Safety Commitments:

  Assess the risks posed by their frontier models or systems across the AI lifecycle, including before deploying that model or system… They should also consider results from internal and external evaluations as appropriate, such as by independent third-party evaluators, their home governments, and other bodies their governments deem appropriate.

  And if they didn’t give pre-deployment access to UK AISI, it’s a fairly safe bet they didn’t provide pre-deployment access to any other external evaluator.

  1. ^

     The violation is also explained, although less clearly, in the Time article:

     The update also stated the use of “third-party external testers,” but did not disclose which ones or whether the U.K. AI Security Institute had been among them—which the letter also cites as a violation of Google’s pledge.

     After previously failing to address a media request for comment on whether it had shared Gemini 2.5 Pro with governments for safety testing...

60 U.K. Law­mak­ers Ac­cuse Google of Break­ing AI Safety Pledge

• Joseph Miller 26 Aug 2025 14:31 UTC

  32 points

  21

  on: Joseph Miller’s Shortform

  Startups often pivot away from their initial idea when they realize that it won’t make money.

  AI safety startups need to not only come up with an idea that makes money AND helps AI safety but also ensure that the safety remains through all future pivots.

  [Crossposted from twitter]

• Joseph Miller 22 Aug 2025 0:59 UTC

  5 points

  0

  in reply to: Arjun Panickssery’s comment on: Ar­jun Pan­ickssery’s Shortform

  In other words:

  $\text{Ought to be done}\subseteq \text{Can be done}\subseteq \text{Actually done}\Rightarrow \text{Ought to be done}\subseteq \text{Actually done}$

  My fuzzy intuition would be to reject $\text{Ought to be done}\subseteq \text{Can be done}$ (step 2 of your argument) if we accept determinism. And my actually philosophical position would be that these types of questions are not very useful and generally downstream of more fundamental confusions.