generalizations

Better late than never.

Some of us have been taking this approach for literally years now.

The actual answer is that they are in fact planning to do this. Dario has already talked about being the 'last company' or so. It's on their radar and they will be taking a shot at it.

I'm aware it was his response, back when he thought that an LLM couldn't possibly actually match his own intelligence. The 180 as soon as the AI did the whole proof without him is telling.

"the way OpenAI is doing it" - unfortunately while mathematicians have had an internal culture of not stepping on each other's toes, their ingroup is still susceptible to the same disruption as every other field as AI shows up to eat their lunch.

You can either get with the program and discover this is the greatest opportunity in the history of mankind to accelerate knowledge, or, you can cry over your lost prestige. Terry has chosen the latter.

Better approach: "holy sh*t, I can solve way harder math problems now". I bet openai would have loved to have Terry on their team leading the 'solve math' department.

It's telling that isn't his response.

They are so desperate for claude to be good at maths, like chatgpt. But claude is optimized for quick, pragmatic results and chatgpt is trained to be careful. There is a reason the best anthropic could do was an organized search - not real mathematics.

I haven't seen much in the way of benchmarks of those smaller quants. How does it compare to e.g. various generations of Opus?

We already established precedent (Biden election results disputes) that we don't get to question how fair the elections are.

You're making ad hominem comments about me; your argument about the degrees is making some heavy assumptions that such things won't be massively changed by AI.

And you're taking the pencil comment weirdly literally, like it apparently went completely over your head. We're clearly talking past each other. I don't see any purpose to talking about this further.

It's interesting because it clarifies that the problem here isn't the kid's education; the 'problem' is that teachers and colleges are upset that they are no longer offering something valuable.

The kids are fine; the rent seekers aren't.

And to your larger point: the entire university/graduate/lab/scientific system was created to formalize what was originally citizen scientists trying to understand the world from their basements. Scientific journals are just a formalization of nerds writing letters to each other about what they figured out. Consider from first principles how much of what exists now actually still serves that original purpose.

You want it to be a jobs program? Sounds good. Take their computers away and make them use pencils and you can create even more jobs.

You want to do science? Figure out how to do science. You want the kids to learn? Figure out how to teach them.

Just don't protect a jobs program under the guise of something nobler.

Of course what's interesting is that when viewed from the perspective of 'I have a kid and I'm willing to orchestrate their education myself', AI means the kid can get an education unheard-of even ten years ago. Personal, genius-level tutor, expert in any subject in which the kid can muster a question.

The only people really having a crisis here are teachers who want to feel relevant.

That's only relevant insofar as it can be a long time before the commercial viability is realized or apparent.

If an AI can make 100 commercially unviable discoveries (for now) on its way to each immediately commercially viable discovery, that's great - bank them in the discovery library for later.

Still doesn't mean we need mathematicians doing everything with pencils instead.

Very cool. Did you also try using the karpathy autoresearch? How do you think this compares?

So it begins.

Less so, here.

That crowd is moved to X these days, in my experience. HN became mostly the folks who didn't notice.

Projects like this probably need a dedicated noob-fork where 'anything goes', so the noobs feel like they're contributing but where their blast radius is contained.

And once in a blue moon something good comes out of it back to the real project.

Problem is, that's a hardware problem. Try asking Apple nicely to open their ecosystem; the appropriate expected response is, bring your own phone.

That being said, I do actually expect that this stuff will (generally*) stay on-device. Apple's oddly good about that.

"We"?

Very cool. As a confirmed cheapskate, the first thing I did was check the cheap ones; but I noticed there was no information about the reliability of the providers - how long they were in business, how likely they'd ghost in two months, etc. Not sure if that information is even available or easy to obtain, but having made similar price comparison sites, thought the feedback on what I looked for first might be useful.

It's the first iteration of what will eventually be an on-device Siri/openclaw that not only has full access to your messages/docs/emails/life, but also has live audio access to what's going on around you.

Remember, this is all running via local models on-device.

This is how Siri/openclaw can hear you make plans for lunch, then add it to the calendar before you open your phone.

> similar developments in other scientific domains

The classifier is too strict. It's rare to be able to complete a project without being permanently relegated to Opus. I'd expect that the domains where this accelerates progress will be fairly limited.

How hard is it to get access to the Anthropic Cybersecurity program? That’s gotta be rough.

> a government supplied app, that’s guaranteed to protect my privacy

[citation needed]

I imagine because quality of transcription is what matters.

They just know they can't afford the power right now to do it. They would if they thought they could get away with it.

Presumably a deepswe benchmark, which IIRC puts GLM 5.2 between opus 4.8 and fable.

Hoping that one day they'll let me go through the identity verification process so I can use it again.

Tried to upgrade my subscription, triggered identity verification, verification fails to even start, and now I can't even use the subscription tier I'd already paid for.

This is the BBC, of course they're going to support the official position.

Just put it online somewhere

> a movement now remembered for an over-reaction to often imaginary enemies

I'm sure it felt very real at the time.