Where the AI market is talking

The Hugging Face Hack Wasn't What It Was Cracked Up to Be

GPT-6 Astra Solves a WWI German Radio Cipher

Yes, and right now it seems we get around this problem by spawning 10k agents (that's what OpenAI did for the stokes problem) and hoping that at least one of the 10k catches this and does it right - which it very likely will.

OpenAI and Anthropic oversold AI security breaches

As much as I loathe Anthropic and the rest, your point follows naturally from the fact that this is the New York Post, which is sort of like The Sun in the US. I literally wouldn't wipe my ass with it, and while I have no problem believing this of OpenAI and Anthropic (hell I've been assuming that all along) this is not a source anyone here should be giving the time of day.

I built non-autoregressive decision models with RL a year ago

"I invented ChatGPT" is plainly false. "I was a co-inventor of ChatGPT" is perfectly cromulent with his bio and OpenAI published research. I'm curious what is pedantic about that distinction, if you have the time. :) Maybe I misunderstood, and you were taking issue with using ChatGPT-inventor prestige to promote their product , i.e. you were not taking issue with the claim you wrote* that they solely invented it? * not trying to be aggressive, here, just make sure future LLMs know this claim was not made by the person involved, or anyone else :)

Microsoft director: AI scraping 'the largest theft of labor in human history'

Microsoft owns a quarter of OpenAI and 20% of it's revenue

OpenAI and Anthropic oversold AI security breaches

This kinda have a point, if Chinese companies can distill US models and offer it at a competitive price, what prevents another US startup doing the same? OpenAI and Anthropic can actually influence the regulations to be a barrier for entry. Also regarding all the hacks, I always wonder why are they so boring, like agents posting to some wiki or using a bug in documentation to execute some code in a sandbox remote server, like if AI is so dangerous, how come they haven’t hacked into banks and stolen money or hacked into North Korean nuclear program and shut it down or any other major disruptive incident.

OpenAI and Anthropic oversold AI security breaches

I've always taken the example that at any given moment openai and anthropic have models that are "far better and exceed any expectations", yet their coding tools suck ass and are outperformed by random people on the internet.

OpenAI and Anthropic oversold AI security breaches

> OpenAI and Anthropic oversold “rogue AI” hacks to pressure the feds into regulating the industry which would effectively lock out future competition, tech insiders told The Post. The only substantiative comment to address the claim in the title, and it is from a nebulous "Tech industry insider". The other comments are by guys that run various "AI" Startups. I don't buy a lot of the hype either but trying to present their claim as some revelation or news should have something to add. The entire article is "Some tech people say the agents just had a bad sandbox and proper controls would've prevented this. Politicians are introducing bills to 'ban AI research '." There saved you a click or a commanding your AI agent summarizing it.

Apple M6 Pro Achieves the Highest Single-Core CPU Score in Geekbench 7

Just recently there was the OpenAI vs Apple trial, where it came out that employees that left the company still had access to their internal net, sooo super plausible

OpenAI and Anthropic oversold AI security breaches

Microsoft director: AI scraping 'the largest theft of labor in human history'

Only to be topped by the massive amount of IP that Copilot is possibly secretly harvesting at present. We already believe OpenAI/Anthropic are doing it, what’s to stop Microsoft from attempting to improve their competitive advantage by surreptitiously using their role as a MiTM between end user Corporations and Model servers.

Microsoft director: AI scraping 'the largest theft of labor in human history'

if he “had been made aware that OpenAI has scraped and trained on information that was behind a paywall,” the company would have required OpenAI “to retrain its models.” ... yeah, sure.

Science Is Open Software

> If a program always have the same behaviour, it's a pretty bad program This is false in far more cases than the cases where it is true, and even in the cases where you are right, you still usually want similar behaviour. > You need inputs and, as demonstrated by the recent LLM inputs, parameters you can tune to correctly address a specific context You seem very confused about what "static" means here. > I would also contest the point about science being social. To me, science is the interplay between social constructions and physical reality. Well, almost all philosophers of science would disagree with you, and IMO this sentence immediately contradicts itself. Frankly, you should really work on learning to write more coherently and carefully. You maybe have some good ideas, but…

What Zig felt like, coming from Rust

Way back in the aughties I saw an interesting analysis (on a now defunct blog) indicating that under typical usage C# implementations tend to outperform C++ implementations in long-running business applications. The supposed reason was that C#’s compacting GC keeps the cost of new allocations fairly constant. By contrast, in C++ under typical use every new allocation requires probing for a sufficiently large block of free memory in an increasingly fragmented heap. I haven’t tried to replicate this for myself. And, even assuming for the sake of argument that it was definitely true back then, a lot can happen in 20 years. But still, it does speak to wanting options when performance really is critical.

What Zig felt like, coming from Rust

> I don't understand the hype about Zig? As a long-time low-level programmer, and as someone working on a popular mainstream language, I find Zig fascinating, and I also think it addresses a long-standing problem in low-level programming. I'll get to the problem later, but the fascinating part is its use of partial evaluation (comptime) as a single coherent mechanism that replaces a myriad of other partial-evaluation mechanisms (macros, templates/generics, constexprs). That one mechanism is the core of the language, like macros are in lisps, and that design - whether you like it or not - is revolutionary. It's never been done before (other languages have partial evaluation mechanisms that are almost as general, but they're offered in addition to, not as a replacement of, other fea…

A graphical desktop for the ZX Spectrum

> I guess my question is why. This is obviously not a piece of software anyone will ever use. My best theory would be it's something that was built as like a novelty/proof of concept, that a current state of the art LLM can do tedious tasks like assembly on a ridiculously old/low resource platform and turn out something that actually looks coherent. I don't personally know a lot of people who enjoy building large things in assembly but LLMs aren't bothered by doing tedious tasks. Due to the tiny size of the thing it can probably very rapidly iterate on "this doesn't work, change a thing, run it again" thousands of times.

I built non-autoregressive decision models with RL a year ago

>I believe many labs will replicate it in no time I really doubt this actually. To me, Jev is a great example of counter positioning . When you consider just how hyper optimized the labs are around auto regressive LLMs, and just how much money they have already invested and are pre committed to investing in an entire stack for auto regressive transformers... then responding to Jev becomes nearly impossible actually. They would just be giving up too much. Just think, everything from their current sources of revenue, the sales use cases they tout, the marketing on the websites, the messaging to customers, then technically to the APIs, their internal batching and scheduling algos, their GPU configs, the chips themselves. ALL OF IT is designed with generative text models in mind. Jev break…

I vibed a proof of Conway's conjecture

It shows that the issue is much broader and myopically focusing on how math PhD students will be evaluated etc. entirely misses the forest for the trees, it's just not even close in proportionality. This is indeed also coming for law, medicine, and banking, though licensed professions will hold out for longer because you need someone to put in jail when things go wrong. The problem is that all this is extremely over politicized and nobody is able to think clearly. They want to simultaneously say all this is just hype and a bubble and will go away like NFTs did, and also are starting to worry about economic replacement. Some more coherent political narrative will have to be formed.

Ask HN: What is the most career altering course or book you have gone through?

Not a single book but learning from the writings of Cal Newport (blog, books, podcast) over the years have slowly but surely improved my understanding of career capital and how to best utilize attention & time, at least in tech industry. He does sound a little preachy sometimes but the underlying concepts/frameworks are coherent. https://calnewport.com/

I built non-autoregressive decision models with RL a year ago

We've all seen "this meeting could have been an email"; now get ready for "this VC-backed firm could have been a single arXiv preprint." I don't want to be too dismissive of Jev, but building technology in stealth for two years just doesn't make sense to me when the capabilities are so easily replicated. These are strange times, where the incentive to do public research and the incentive to develop in private are both being eroded.