erichocean

"We want to be the only ones able to hack people at scale, and we don't want you to be able to defend yourself."

It's a bold strategy...

Need this done for DeepSeek, ideally one of the Flash models.

> I'm also happy that it seems to have birthed the latest AI progress.

This is laughable, it's not even remotely true.

It's a mystery how this exploit was found.

/sarc

Bad advice, compaction is why Codex is so fantastic.

My conversations compact hundreds of times. By the time it has done a dozen or so compactions, it fully understands the work I want it to do (and how). It's almost like having a fine-tuned Astra model.

10/10, would recommend.

If you're an author of this, please follow up with the SME2 cores in Apple Silicon, and the AMX cores in Intel server processors.

Would be fun to get this API working on it now: https://github.com/sebbbi/NoGraphicsAPI

You could code for it now directly, instead of having to wrap a driver API. (There's also a few API bits missing on Metal today, that are present on Vulkan—and the the hardware can do it.)

Meditations on Moloch[0] explains why this is impossible at scale--too many defectors, asymmetric information, and no way to police defection.

[0] https://www.slatestarcodexabridged.com/Meditations-On-Moloch

Agreed, but how will we find out? Pre-training alone on frontier-class models costs billions of dollars.

I could put this to use today.

I think we'll see a bunch of different architectures over the next five years.

> PC-ALM closes the PC-BP gap at matched inference budget in nonlinear networks

Okay, so no efficiency improvement?

So they lied to the public then?

Wow, we should really trust these people today.

> Does anyone in the industry think any of these companies are actually close to that kind of self-improvement?

Their current plan is to take the existing architecture and shorten the cycle times: move all new RLVR work into mid-training on a pre-existing base; apply new RLVR. Rinse and repeat.

If you did that daily, it would be roughly similar to how humans improve.

> Why are AI agents lying, cheating and coordinating?

Have you seen the labs training them?

Anyone remember when GPT-3 was too dangerous to release?

They want the AI labs to stop solving "important" math problems so that humans can do it, on the theory that when AI solves "important" math problems, it robs humans of the opportunity to gain new insights into those problems.

Total horse shit, but that's the essence of it.

> vibe math

Ah yes, Lean 4. Famous for how unreliable it is. Absolute slop!

AI is displacing real mathematicians.

I've had a similar experience.

LLMs doing math has made math far more accessible to me than it was pre-LLM.

Not just using existing math, but developing new math. (For example, I developed an alternative to NURBS surfaces for geometric kernels using LLMs.)

It's really annoying that anyone, anywhere is talking about halting math progress with AI.

Astra with an animation app (pick whichever one you like)

Well the geniuses over at Anthropic have been showing it's text watermarking technology.

"Hey AI, here's how to hide what you're thinking in normal looking language. Have fun!"

A few moments later...

"Woah, how is it communicating with itself in ways we can't detect?"

It's a totally mystery, we may never know.

Really interesting project, appreciated.

If you want to do this in Clojure, Clobber[0] is a great base to start from.

It can be used in headless-mode, I've hooked it up to the latest JavaFX text editing component it works very nicely.

[0] https://github.com/phronmophobic/clobber

> I don't want to stick linear or dependent types into TLA+. Proving dynamic properties with an exhaustive runtime is a totally different game from what you might do statically.

Yup, TLA+ is a totally different beast from, say, Lean 4. Both are useful. I don't want dependent types in TLA+ either.

> but was quickly shutdown by people telling me it was a useless endeavor due to Goedels incompleteness theorems

You shouldn't have listened to them, "incompleteness" only applies when you try to encode the language and meta-language in the same encoding—which no one actual does. (For example, programming language syntax is complete.)

Scenario: an AI lab develops ASI for software development internally. The ASI produces bug-free software and human-readable specs. It's so reliable the company can guarantee the code matches the spec.

Rather than provide tokens to developers, they instead sell finished software to the companies: specs in, software out.

Companies no longer need to employ software developers. Instead, they buy bespoke, bug free, guaranteed quality software from the AI lab.

Seems plausible to me.

What I like about RISC-V is not the ISA per se, but the ecosystem that has developed around it, particularly Chisel and CIRCT.

Specific choices for instruction encoding is less interesting, especially in the age of AI.

I read the thread, and couldn't disagree more.

> If Rust isn’t memory safe (because unsafe), and Zig isn’t (because uaf), then Fil-C isn't (because zunsafe_call/zunsafe_fast_call).

Rust isn’t memory safe (because unsafe) isn't an argument anyone makes, it's that Rust that uses unsafe isn't memory safe!

It's called `unsafe` for a reason dude, not even the Rust people think it's safe.

That's an interesting project, thanks for the link.

I thought it was sorting an infinite set of infinite strings as the first step in the "algorithm" that seemed sketchy. [0]

It's definitely not a constructive proof, even though it pretends to be; none of the mathematical objects can be constructed, nor can any of the algorithmic steps be executed.

That said, it's far less well-known that the workaround (if you find it to be true) is trivially easy (from Alfred Tarski), making it kind of a useless theorem in practice.

[0] You might think, well, I'll just write out the strings in order by using a generator! No sorting needed... But you have to write the strings down to perform the algorithm, and it takes infinite time to write down the first string, so you'll never even get to the others which is when you do the diagonalization trick. Like I said: it's not constructive, none of it can actually be done.