Re: Seems some are getting confused…
Liam your “explosion” is actually very informative and aligns more with my overnight thinking about BSE and the discovery of the extent to which beef products had permeated the vegetarian food chain and that strict vegans need to check E numbers…
> “C is crap. C was crap in 1969 and it is worse now.”
Coming to C from a background in the “classics” (ie. Algol-68, Pascal, COBOL, Fortran, ADA), Prolog and various Assembler languages, I was surprised just how much people raved about C. However, as that is what the world seemed to want, developing a C workbench (think forerunner to VisualStudio) for the MSDOS PC seemed reasonable…
A concern has been whether replacements eg. Rust - influenced by C, are going to be significantly better.
A second concern has to be the amount of C source being used for LLM training, which can surely only perpetuate the use of C.
>” The thing here is that right now, bots can find bugs. Whether they can _fix_ them well is not yet clear but by Jove they can find them faster than ever. So objecting to bots on safety grounds -- the nut allergy comparison -- does not stand.”
There is a lot in here. Firstly totally agree using the best tools available to find bugs is a good thing and if the best tools are AI then let’s use them. Your second point, the “fixing”, is the real issue, like you I have doubts and it is here we need to be most cautious about the uninformed use of LLM based AI code tools. It is this context, where LLM AI can introduce unexpected code/side-effects, I think the parallels with nuts and BSE are relevant. However, these side effects still don’t necessarily mean the resulting code contains “AI features” which users would recognise as AI augmentation.