AI vs AI Review

Format: Have AI A write code, have AI B review it, and you judge who is right.

What you need to judge:

• Are the issues AI B found real?
• Did AI B miss anything?
• Did AI B "over-review" (flag things that aren't actually problems)?

Reasonableness Assessment Guide:

• "Suggest using parameterized queries to prevent SQL injection" → Reasonable
• "Variable names should use camelCase" → Possibly reasonable but not critical
• "Suggest rewriting with CQRS pattern" → Possibly over-engineering