Is there anyone who would like to research hallucinations in AI and ways to reduce costs?

These days, with the development of the AI era, it seems there is a bottleneck
Even if there are so many AI-related developers
It seems like bottleneck issues, hallucination problems, and costs are unresolved issues
Is there anyone who would like to do research together?
In my own way, to solve this problem, I
I did research on the library and uploaded it to GitHub
Looking for someone to conduct research together

[ Draft Answer ]
|
v
[ Claim -> Attack -> Evidence ]
|
v
[ Judge Verdict ]
| pass | patch/rewrite
v v
[ Ship ] [ Retry with feedback ]