AI outputs can sometimes show contrarian or biased behavior that users don't expect. A tool could analyze text outputs and highlight potential biases.
Target researchers and companies using AI for sensitive applications.
Start by detecting obvious patterns, then refine with machine learning.
The risk is that bias detection itself can be subjective.