Neural Network Integration

[ritcat14]

Applied neural network to core thread handling. Deleted large sections of core functionality in direction of efficiency, works but missing thread locking as AI decided to use its own management system to handle threads. Locking system needs re implementing and AI needs access to locking method restricted.

[ritcat14]

Neural Network game access should be limited to:

• Thread Management, but not the ability to effect current locking system. Should only be able to control Thread execution order for most efficient execution
• Resource loading/saving and data processing

Need to revoke access to writing code in areas of the project, some sort of code block (comments?) that it can read to know if it can or can't overwrite the section

[ritcat14]

Neural Network reaches point of "max efficiency" when efficiency coefficient reaches a point it cannot get any lower, but does not reach 0. At this point, it appears the AI says to itself the only way to lower the value any more is to start deleting random sections of code. This causes it to delete itself and core functionality. Even code comments telling it sections that are "out of bounds" it overwrites as it has nothing left to do, so rewrites itself to be able to delete those sections. Options are to limit AI based on efficiency coefficient, limit what can and can't be deleted (i.e physically revoke access to core features) or have it become idle upon completing task, preventing it from deleting further code.

[ritcat14]

Realised AI does not have an understanding of the purpose of the code it is making "more efficient" (i.e faster) and thus deletes code to make the method calls faster. Issue related to above but not sole cause.