How do you propose to avoid "distortions of AI" in your mega-examination?

... well then, why not use those same protections (proctoring, monitoring, auditing) in continuous examination?