j_maffe

I highly doubt given OAI's latest streak that the models didn't know what they were doing.

I'm not sure I agree. More is different. Being able to execute logic at a higher scale and speed would make some previously infeasible intelligence tasks possible, resulting in a new level of intelligence.

That would be the best case scenario. I honestly wish things would finally slow down a bit. I don't see it happening.

Would you argue the same for allowing everyone to have guns?

Yeah the Chinese fear-mongering falls a bit flat when coming from a point of maintaining US supremacy

> in a fraction of the time

Well if you do the math, the number of agent-compute time in total, given the insane number of agents thrown at the problem, might end up being comparable in time, if not for the budget.

I think if you use an LLM just to proofread then it'll not be able to insert a strong enough watermark.

The tasks are the thing to really look at here:

https://github.com/harbor-framework/terminal-bench-science/t...

> I worry this doesn’t check correctness

Then it's not a valid benchmark. I agree though they're not reliable enough to just put results in a paper.

The Purpose of a System is What it Does.

No but a provider with more amicable terms can.

Anyone has a link to a report of its capabilities? I can't find a reliable source.

Not if there's a ZDR policy.

I think the idea of having a human-written persistent document describing the operation of the code is a great idea. This document acts as the prompting interface instead of the chat window and changes can still be tracked. Surely something as simple as a skill.md can be made for such a setup, right? I think the pseudocode style is a seperate axis to this setup.

They would rather ban to deterr other governments from taking such measures.

> “It’s no secret, I disagree with the prime minister," he said. “I think targeted assassinations should be carried out in Gaza, taking down 30 to 40 every night,” said the far-right Israeli lawmaker.

> “Not just those who pose an immediate threat, there are people there who are not worthy of life. They shouldn’t live. They’re not even people,” added Ben-Gvir.

Not sure how the full quote is supposed to help.

Almost all of my karma is from a single post link that blew up randomly. I think this is the case for most users' karma here.

But that seems to have little to do with the quote... I'm confused.

You think they'll shutdown FB?

What does cheating benchmarks have to do with breaching financial user agreements? To use your argument: all it takes is one whistleblower to get the company sued for billions of dollars.

> If not, congrats, you're on the same intellectual level as a racist.

I remember HN comments used to be of higher quality than this.

> So I'd rather spare myself from explaining it.

Being snarky doesn't really help your position.

It's good as a community to communicate what we prefer to read. At least that way the author knows why they're getting the cold shoulder.

Such a brain-dead idea...

Visualization of the configuration: https://claude.ai/public/artifacts/9db65255-16ff-4f8e-8be1-1...

Cheating how? The proof is in the pudding...

> No one is claiming that time is a random variable drawn from a normal distribution.

You are doing that implicitly by fitting a Gaussian curve.

> statistically speaking

That's a very big word you're using there for what is basically making shapes out of clouds. A bell-curve is the amortised function of a random variable with a mean and standar deviation. What does that have to do with a timeseries dataset?

I'd call it significant that the number of questions halved within one year following the release of ChatGPT, the biggest relative or absolute rate of decrease in the timeseries.

You don't just fit a Gaussian distribution to a timeseries dataset. That's not what a Gaussian curve is designed for at all. https://www.explainxkcd.com/wiki/index.php/1725:_Linear_Regr...