AI models for finding pointers or AoB?

[ersatz]

I see a ton of effort going into computer vision for analyzing captured gameplay to us AI to build aimbots. But I see almost nothing in regards to using AI models to point scan and auto calculate variable integers for games. I'd assume an AI could pretty much auto complete a table of pointers related to what is seen on screen for example in the same way yolo, or rootkit do for analyzing friendlies and enemies. The same logic could be used to detect on screen numbers as a way to start.

AI could start auto scanning for anything it sees on screen and then play with altering the pointers to see an on screen change that matches what it set the value to. I don't see why this couldn't be pursued.

Does anyone know or have seen deep learning projects that are working on this?

[LeFiXER]

By and large computers are binary machines which are just 1s and 0s. Humans have turned this language of 1s and 0s into a human-readable format with the advent of the assembly language. From there higher-level languages were developed such as C, C++, Java and so on. It's only with those languages that the numbers were confined to set sizes we defined with each type. Numbers are all hexadecimal in memory therefore you quite often see values within a memory region multiple times. It is quite easy for scanners to mistakenly read a value as the value needed because it is in memory multiple times hence why you see thousands, if not millions of results when you scan for a simple value. I just don't see the feasibility of this, perhaps I will be proven wrong, perhaps not - only time will tell.