0 minutes because the entire 11 years this project has been open source on a dosbox branch; therefore, it has all been harvested in the training set, and it's a matter of requesting that the model reproduce the training set and replicate my and others' memory debugging adventures.
Reading my comment again, I realize it sounded not okay and even rude, sorry about that! I didn't want to say AI can just replace 8+ years of hard work and probing memory by hand. I'm curious what parts of the reverse engineering process can be faster with modern AI tools today? What parts still need 100% human logic and trial and error. Amazing job for not giving up on this for so long!
[–] daniel_rh 1 day ago
[–] atakmim 1 day ago