As suspected, this project was possible, in part, because of LLM models. I have also been exploring in using LLMs for Gameboy Color testing.
There is something quite ironic about old hardware becoming increasingly useful due to current developments in AI research. I wonder about the repercussions of this.
Before anyone gets the impression that the whole thing was done by AI (like I did), it seems this wasn't vibe coded, more AI-assisted:
"Generative AI (Large Language Models) have been used for research and debugging help in the course of this project. Small amounts of boilerplate code have also been written by LLM (C++ class and interface definitions). I do not intend to make this a "Vibe Coded" project. Pull requests automatically generated by a LLM tool will not be accepted."
As AI companies buy up the supply of hardware, we'll be increasingly dependent on obsolete hardware. So it's a good solution for the problem it created.
There is something quite ironic about old hardware becoming increasingly useful due to current developments in AI research. I wonder about the repercussions of this.
"Generative AI (Large Language Models) have been used for research and debugging help in the course of this project. Small amounts of boilerplate code have also been written by LLM (C++ class and interface definitions). I do not intend to make this a "Vibe Coded" project. Pull requests automatically generated by a LLM tool will not be accepted."