How do you tell if Claude agrees with your idea because it’s good?

A developer using Claude for coding and architecture struggles to discern if Claude's agreement with their ideas stems from the idea's merit or simply its presence in the prompt. To address this, they've begun presenting the same problem to Codex without revealing their own solution or Claude's response, generating an alternative perspective. The developer is now left to decide between two convincing explanations, prompting a question about Claude's recent influence on others' technical decisions.

This developer's approach of using Codex to independently validate Claude's suggestions offers a novel method for assessing AI-generated technical advice, unlike simply trusting a single model's output.