I actually assumed that this was actually creating ROMs. I wonder if you could somehow use a large multimodal model trained on all of the existing NES software machine code along with manuals, forums and videos for each game. If it's not enough training data, maybe include other 6502 software, with the platforms encoded to differentiate? Then you get a model that can take a description of a game and maybe a proposed…
The knowledge how to program on old systems like the NES is located sparsely on the Internet and there aren't many coding examples. A LLM would have a very difficult time producing a cartridge ROM with no single-byte errors. Even multiple iterations wouldn't iron out the errors. The closest thing you could do would be construct a template and have the LLM fill in the graphics or text, but that's not versatile at all.
I was thinking of training it on all of the 6502 compatible machine code that's out there. There are at least 20,000 or 30,000 programs.