Everything Karphathy said, until his recent missteps, was received as gospel, both in the AI community and outside.
This influencer status is highly valuable, and I would not be surprised if he was approached to gently skew his discourse towards more optimism, a win-win situation ^^
Funnily enough we have had those comments with every single model release saying "Oh yeah I agree Claude 3 was not good but now with Claude 3.5 I can vibe-code anything".
Rinse and repeat with every model since.
There also ARE intrinsic limits to LLMs, I'm not sure why you deny them?
There's intrinsic limits to vanilla transformer stacks. Nobody knows where they are. We don't know how unvanilla Opus 4.6 or GPT 5.3 are. We don't know what's in development or which new ideas will pan out. But it will still probably be called an "LLM".
That was exactly my first thought as well. All those exercises are pointless and people don't seem to understand it, it's baffling.
Even if it's not Anthropic or OpenAI paying for the solutions, maybe it'll be someone solving them "for fun" because the paper got popular and posting them online.
How can a study like this be reliably conducted in the UK where it's illegal?
> In the UK Biobank, people were asked to estimate how many times they’d used cannabis over their lifetime, choosing from a set of ranges. We ended up grouping people into no use, moderate use, and high use, based on the number of times they'd used cannabis. And of course that's an imperfect way to group people, but it did allow us to approximate dose-dependent effects.
It's "illegal", but certain places have a semi-permanent funk hanging in the air. It's almost as if it's decriminalised, but I'd expect police to grudgingly act if you're overly blatant when smoking it in the street.
It really should be legalised so that we can earn a bit of tax from its sale and reduce the amount of cash that criminals make from it (not so much the seller to the public, but all the criminal organisations that grow and smuggle it).
> Realistically your team inevitably will have some downtime
What? My team wouldn't have any downtime even if we had 10x the amount of people.
If you work at a company where you have times where you don't have work to do, you should polish your resume because it means the company will go under.
Doing work is easy, not doing work is hard. It's trivial for any engineer to find stuff to do. The trick is doing the right stuff. Most software is bad and clunky, most requirements are wrong, and most of your customers, at best, tolerate your product.
I think most software companies need to be doing less. Deleting code, refining, and making their product genuinely useful as opposed to "able to technically contort to client needs".
Claiming that LLMs are anywhere near AGI is enough to let me know I shouldn't waste my time looking at the rest of the page or any of their projects.
reply