solidasparagus

Is doesn't even make sense, the limiting factor for Anthropic is GPU count, not profitability. They want more adoption at fixed GPU count which demands higher token efficiency.

AI native shouldn't imply idiotic. And designing your execution approach around a strategy based on a technology that doesn't exist is idiotic.

I don't think it's profit, I think it is genuine functionality. I think that they build features for their internal users while having no token concerns. And this mindset of using LLMs to solve product problems leads to useful features (e.g. agentic auto-approval). Then there is a bit of an issue when they try to ship a genuinely useful feature but now token cost matters.

Why is an LLVM release any different than a frontier model release? Well known, concrete issues get identified, evaluated against, RLVR'd on, and a new release comes out that doesn't have that issue. Or you can do that process yourself on an open-source model if you have the resources.

LLMs are not magic things that are out of your control, they are simply complicated. But there are concrete issues that are identified, people sit down and decide to fix it, build evals, run ablations, and they produce a model without that deficiency. It happens every day.

You look at the inputs to the model at the trace level to root cause context construction vs model deficiency. You collect examples of the failure cases, build eval sets, and run ablations. That's black box debugging, but systematic debugging nonetheless.

Not 1 in 1000 software engineers has the ability to debug a compiler. And you can absolutely debug agent failures, but similarly almost no does. In both cases for the mast majority of engineers, they are black boxes.

I'm not sure. That is also my experience, but I'm not completely sure that isn't my fault and the compiler comparison would be totally valid if I just used the tool better. When I actually write fully fleshed out specifications and have the LLM handle the implementation, I rarely encounter bugs.

I don't think I've seen a Sol 5.6 or later model fail at a task with a verifiable outcome.

It's when I try to do too little of the defining that I get LLM bugs. In which case maybe programming hasn't changed that much, LLMs really can be reliable compilers, and we're all just being lazy.

Neither are compilers? But you still trust your JIT compiler. Determinism is unimportant, likelihood of correctness is what matters. Plus agents are deterministic, no?

How does it differ other than that people aren't currently choosing to write clear, well-defined high-level specs and feeding that into their agent?

Complete comprehension has never been a thing. Unless you were working solo with no dependencies, you never completely understood what was being shipped. I don't think an npm import vs an LLM implementation of a well-scoped, well-abstracted library is any different.

At agentic coding velocity, tech debt becomes obvious much more quickly. Empirical evidence is a bit of a ridiculous bar to hold for software development - we are as quantitatively blind now as we have ever been - but at least for me the anecdotal evidence is very clear. Today's agents hill-climb and their software architecture is pretty reliably bad without a human in the loop.

And it's debatable how much better they will get quickly. Verifiable rewards for maintainability of software is, I think, possible but very hard.

0.5 tok/s I think is their claim. Basically skynet

You should read the report, it is interesting technically if you ignore the fear-mongering. It distributes the model's layers over multiple compromised devices.

This technology has surpassed the the world's best mathematicians at the (explicitly and openly stated) goal by which they measure and reward mathematical progress. Such an advancement that it seems to have thrown the entire field into existential self-doubt. And then someone sits at their keyboard and says it is just a fuzzer.

I think the answer is literally no, they don't have the ability.

Every option is going to lead to someone shitting on OpenAI for what seems to be a pretty huge accomplishment. There have been opinions written by some mathematicians that OpenAI should just share the work that they have now so that people who are working on any solved problems can know. Which seems reasonable to me.

Is mathematics research even a job you pursue if survival is the question?

But I don't think OpenAI intended to hack HF and I doubt you'd be able to convince people they did. It's such a weird story to try to tell in a court of law where this was intentional. To what end?

> Even for candidates for whom my feedback was scathing, they are still generally positive about the experience

If you have scathing feedback for someone you are interviewing, you should probably just shut up. Or at least deliver it in a different context where they aren't being held hostage and having their every reaction scrutinized. Sounds miserable, I seriously doubt they were actually "generally positive"

The cheap, fast and smart-enough LLM space has been wildly neglected. Jev is one of the few players truly targeting that space. And for a lot of people it is the first time they are asking "what could I build if llms were interaction-speed fast?". The answers are cool, the problem is that Jev is not, I think, smart-enough yet to have that many applications, but it's smart enough that you can start to see what they will look like.

I don't think you need to trust them if you are running on something like bedrock. Then I think you're trusting AWS or whatever cloud equivalent you want.

Why does this necessitate using closed models? I don't see the difference between putting your data on a cloud DB and using a cloud model.

Did you try it? I'm curious if "certain ML accelerators" is an anti-China restriction.

This is an Apache 2.0 open source project

Don't you end up sitting there waiting while the agent goes off and executes the plan?

I had the same experience for my writing that was AI generated.

I am being specific about cyber crime in the united states, which (if AI is to be believed) requires intent.

And for negligence, there's a bit of an issue where the HF hack was probably more scary than damaging. HF is asking for compute as repayment instead, which would be fun because it completes the OpenAI -> HuggingFace -> NVIDA -> OpenAI cycle.

They can raise cash by selling off the stock they have spent a decade buying back.

I'm not a lawyer but my understanding is that OpenAI didn't obviously break a cyber law because it requires intent which doesn't exist (with the definition of intent in a case like this being an open grey area). I think rules around making organizations obviously legally responsible for the cyber attacks performed by agents under their control would be good.

I think they are genuine believers, but I think it's naive to not think that commercial pressure doesn't play a role in their positions, either explicitly or, perhaps more likely, subliminally.

The x-risk stance and commercial stance have evolved to be the same thing - Anthropic must win, and then everything else seems to work backwards from that. Can you believe it, the path they think is best for x-risk involves them becoming filthy rich. And threats to their commercial dominance like distillation get framed in a way to turn them into x-risk concerns.