Some food for thought.
One of the speculated ideas of how GPT-4 works is as a Mixture Of Experts (MOE) (blog)
Following along those lines it could be seen that to add new experts to GPT-4 they are tried out first as an alpha or beta.
One such current alpha is the Code Interpreter (ref) but don’t think of it as something that generates Python code, think of it as an expert that has some abilities to do what some of the other experts can not do such as math or data analysis.
For details on this line of thought see
So a similar expert based on Prolog for other forms of questions/problems would be in the purview of what you seek.
If you expand this out a bit, then s(CASP), constraints, etc. should also be considered.
Note: Code Interpreter is considered a ChatGPT plugin, I also created a ChatGPT SWI-Prolog plugin as a proof of concept. (GitHub) (Details/Demo)