To Grok D

[FeepingCreature]

On Tuesday, 29 October 2024 at 22:07:07 UTC, Bastiaan Veelo wrote:
> Recently, I started using Grok 2 (beta), the LLM that comes 
> with a Twitter Premium subscription, to have it explain things 
> in D. I am not an AI enthusiast and don't know how it compares 
> to the competition, but I find that it does a surprisingly good 
> job. When I suspect that the response is not compleet I can dig 
> deeper and usually it confirms my suggestions, with examples. I 
> haven't caught it hallucinating yet.
>
> So I asked it how it learned D. Point 7 below surprised me. 
> Apparently I then became a bit too personal so it lowered the 
> level of formality and started making jokes :-)
>
> -- Bastiaan.

Note: As AIs are not "conscious" in any sense while training, all 
of this is probably invented. Remember that they are very good at 
creating *self-consistent* explanations.