On 19/3/23 19:10, tomas@tuxteam.de wrote:
I have actually paid for a subscription and have used it for a month now in generating bash scripts and nginx configs. It's mostly pretty good at that.[...] Yes, it is just a simulation of knowledge (it can be pretty convincing at that,though). In other words: if you want an answer from it, you have to know the answer beforehand.
The V4 is better than the V3.5But I usually know what the general answer should be in most cases. The problem is often it will not quite understand my question and give a response to a question I didn't ask.
In this case of the /24 it gave an answer I expected. I imagine it will take a trawl of the RFC and then of actual implementations to find out for sure.
The best description of the AI is it is informative but not authorative.
- Jeremy (Lists)