Artificial intelligent assistant

How can one trust geometric proofs if humans are susceptible to optical illusions? How can one make a proof that doesn't consist of a bunch of symbolic manipulations "formal"?

We don't fully trust picture proofs. Any rigorous proof can be expressed as a bunch of symbolic manipulations. Some authors have even avoided drawing pictures to make this point.

It's usually far more enlightening to visualize the ideas, though. On that note you might be interested in reading the "Lockhart's Lament" essay.

xcX3v84RxoQ-4GxG32940ukFUIEgYdPy b019ef8f7cdcff498d44452c00bfc8f3