folderquestion

If I were to sell the power of LLMs as powerful research agents, and if I had enough money, I could think about introducing little "gems" into the training set of LLM so that my model would be able to discover new theorems and proofs. There is a lot of money at the table, and I am sure there are a lot of genius people with little pay. Perhaps this kind of thinking is wrong?, only bad people would think like this?, how could one detect such a trick without knowing the training set?

I think in little time LLM will have the capability to translate easily code from one computer language to another, for example the LLM could "think" in prolog + lisp instead python. LLM will be able to write thousands of lines of APL or J in just a minute, we will have to accept that python is not the language for future LLMs. Perhaps the power of macros will not be defeated by its complexity because for an LLM a thousand line macro is just another simple code, prolog unification could mean building slots to construct new powerful artifacts that we are not able to have in our minds at the moment. LLM will no longer learn natural language or our most used computer languages, the next step is designing languages for a system that can keep in his head millions lines of code.