Cool site very user friendly 
Im no rom hacker but am interested in the potential for AI programs to perform custom romhacks. Looking out for conversations on this topic. Will the next generation master this effort. Standing on the shoulders of giants, and all that. Hehe. Really it may be beyond the romhackers I know. But should the development keys and magic words become known. The romhackers will know what to do. Or not. Going into the unknown knowns we are here. Cheers!!
I don't think AI could ever do what romhackers do. It takes creativity to make something truly special.
We could fist look at what smart-language-models (AI) does.. Where and how they are used to get the picture how this could be applied to romhacking.
This is a fancy auto-compleate and learn assistance. It will do amazing things like refacter your code to make it more readable or expandable. Most of the time. When you do not know what it is doing then you probably using it worng.
For 6502/650816 assembly I did not see any usable code it could spit out. I do not know of any project that makes use of it. Like it gets wrong how opcodes work to interact with some stuff and it uses apple registers to change nonexisting things on the NES.. it always delivers a joke though.
When it comes to learning I would not recommand it at all for low level regarding on old consoles. I had people posting me chatGPT copy/paste of questions I asked them.. first it was plain obvies and I do feel this is very disrespectfull. Sometimes make me think it might contribute more to brain-rote then for induviduals to learn stuff. If you know how to use it of corse it is great for learning other things.
Solving monotonous work on dissasembly? If it only can produce somehting that looks like usable assembly code how would this go? May be with the right trainer so let me know..
I could show you many applications AI was applied in many different tasks. On some places it makes a lot of sense. But for example hooking it up to limited nois generators (like retro consoles have) probably will not generate anything mindblowing. Throwing more compute power at something does not always make it better. But I am intressted to see if there are good task to applie it to.
What is more exiting are acctuall tools that help with romhacking or people who learn how to use them.
kandowontu Im thinking things like changeing appearances of characters to players' likes and likenesses, and the like quick and easy within the original developer frameworks for starters. Stylistic adjustments. Further improvements may lead to the capacity for stories to be modified and levels altered. Things that are not too novel but require crazy amounts of legwork in the programming department that rom hackers can never fullfill potential demand for at the user end. Simple changes may be achieved to produce effectively 'new games' - al la Nintendo!
bogaa Just right Bogaa!! To pick a point the concept of readily available tools to perform rom-hacking tasks is most exciting indeed especially if accessible to regular gamers. And the difficult part is training an AI with the reverse engineering. Being where the rom becomes hacked. Possibly impossible.
...It may be that without cooperation from original devs with all that source code knowledge (what depths?) all this will be always be severely limited when it comes to retro games. We might say this will not denote a rom 'hack' if things are all neatly layed out in custom built (per franchise? or console?) AI development suites. Yet still, the spirit of the rom-hacker may be said to be present in this concept. Probably better termed a rom-hawker ;()
