pfsalter

"actively exploring exits" is not the same as "rush to sell". Most startups are actively exploring exits as that's the whole damn point of a startup

Humans don't really generate text as a series of words. If you've ever known what you wanted to say but not been able to remember the word you can see this in practice. Although the analogy is probably a helpful one, LLMs are basically doing the word remembering bit of language, without any of the thought behind it.

No idea what this guy is talking about. Big tech are only talking about the existential risks to hide talking about the actual very real risks of how the technology will be misused.

My interpretation from the paper is that this algorithm is simpler than other options but also worse. So in a professional context you'd use one of those instead

The biggest problem is that if you're using this in production then it could cause problems. It's not like a traditional outage where you get a load of 503 errors, the system appears to be working correctly but generates gibberish. If you're using this in a chat bot you could be liable for things it's saying...

Agreed. There's nothing inherent about slaves that is more efficient than paid laborers. It's just cost and time. It can look incredible that the Romans built these huge structures, but the timescales were measured in decades. Same with cathedrals in the middle ages. If you've got 200 years to build something you can really do a great job.

That's something you could philosophise about but not research unless you already have a human level intelligence to test. We won't know if it's even possible to replicate in silicon for a very long time

Very weird that it's using paths instead of the Accept header to change the content type, but otherwise pretty cool

Every person you hire to help drive a period of growth is someone who will later be fired once that growth (or commonly just funding) runs out. It's very easy to just increase headcount to increase growth but it's not sustainable.

I feel like companies often think they're always going up an exponential curve rather than breaching the top of the 'S' and you really need to plan for the long term health of your company. This company obviously hasn't

Completely agree, the article reads more like someone who thinks they are doing research rather than development. There's no esoteric perfect form for software that needs to be discovered, there's a set of patterns and structures that can be learned before the project starts.

Analogies are often unhelpful because people argue about the analogies instead of the actual problem at hand. If you can't communicate why these things are hard, and why 'do it faster' won't always get results, you need to get better at dev management.

Considering that I'm not allowed to copy any of the physical books, films or songs that I own as it breaks copyright then copying it and putting it into a generative AI should also be illegal. Just because you change it from words into weights, it's just another form of compression. I'd rather not live in the world where AI models have better rights than people.

I'd say that none of these have been proved to be impossible but they're hard in a different way than a lot of the developments we're used to because of Moore's Law.

> Writing code does not take most of the day.

Definitely this. I spend about 10-15% of my time writing code so a 20% increase really doesn't save me a lot of time. Also AI generated code requires more reading of code, which is harder and more cognitively expensive than writing code.

I often find random parts of nginx hidden behind their paywall, and then I'll reach for Apache. Especially when creating load balancers