nuc1e0n

Neat. Can this model learn which table indexes to create that would improve performance? In my experience, nested loops will be chosen in a query plan over poorly chosen indexes, but well chosen indexes will result in drastic speed ups.

Is fab process re-engineering something an AI could help with?

Why SDL 2 and not SDL 3? Was the project started before SDL 3 was released?

I mean nul terminated strings aren't great, but neither are strings with a defined length Strings generally have poor performance, so making them easy to use isn't necessarily a good thing.

That's what pascal did back in the day, but 255 byte strings were all that was needed back then so only a byte was needed to store the length. Does that still sound maintainable? Anyhow, some developers put data into strings when they shouldn't, and require doing that in the APIs they publish. Strings, whether NUL terminated or with stored length, aren't always the best choice architecturally so making them easy to use isn't necessarily a good idea.

I'm of the opinion AI slop code will become untenable way before that.

Why bother 'cheating' at all? If you don't want to follow those precepts, then just don't.

I doubt the dropoff is as large as 100k tokens. I start a new session and paste the best results from the previous one as soon as as LLM makes more than a couple of missteps. Theres too much focus on fixing what's wrong rather than going back to what worked and amending in a different way.

If you don't point out what's wrong I find the LLM will go into great technical detail which consumes a lot of tokens, but not 'see the wood for the trees'.

It seems to me human beings also have mechanisms to compact context, which may be why we can forget what we came into a room for when going through doorways. I think it would be interesting to research which markers we use to compartmentalize our thinking.

c memory allocation functions can be implemented in 1000 lines or so. I've done it myself. Maybe more are needed to handle strange operating systems or architectures, as glib does.

Replying to this comment so I can refer to it later.

I disagree. But maximising the acquisition of bits of paper and numbers on a screen above happiness and day-to-day needs is the American way.

The article claims tensorflow uses eigen. That's not entirely accurate, although the pre-packaged builds of tensorflow are probably configured to use it. Tensorflow can also be configured to use BLAS/LAPACK at build time and not eigen.

Maybe you can install homebrew and open source apps to make it more Linux like, but you'll still be stuck with Mac OS's shonky window and task management UI unfortunately.

> but it's much closer to an "interesting lab demo" than an "obituary for human programmers."

Was it being sold as the latter? I thought it was strange that a fuss was being made over this tech demo, and that explains why.

Parsing regular languages into other regular languages is exactly what transformer based LLMs should be good at.

A while ago there was an AI system publisized that was trained to generate animations from the game doom that was sold as AI created computer games. But the output made no sense if you watched for more than a few seconds. Isn't this the same kind of scare tactics dressed up as innovation?