AI code review (Claude, maybe Codex)

[Thibaut]

> Le 8 avr. 2026 à 19:30, David Lang <david at lang.hm> a écrit :
> 
> Thibaut wrote:
> 
>> tl;dr: I don’t think chatbots can fill in for the humans in the loop, but maybe we disagree on that topic.
> 
> I wouldn't think of the AI as a replacement for a human, think of it as you would a static code analysis tool. If it didn't have the AI label, but was just a tool that analyzed the code and identified potential problems, would you have any problem with it?

Yes, for the resource usage side of things, which is pure madness.

> (we've had those tools available in the industry for a decade or two)

Yes and they didn’t require tapping an entire datacenter to run (and weren’t prone to hallucinations, AFAICT).

> I would think of the "AI" as just another version of such tools.

Even ignoring the point I make above, what *you* or *I* think is not the matter IMHO: the matter is public perception. Besides if submitters only got static analysis of their contributions and nothing else, I don’t think they would be much happier either.

I’ll rest my case, others have been more eloquent than I.
 
Best,
T.