Wednesday 29 July 2020

Regulating AI: A practical approach to bias

Here's an automated approach [idea] to dealing with bias in AI. 

Regulatory bodies apply an approach similar to GAN to test bias in an AI system. 

The regulation system presents the AI under inspection with a spectrum of inputs that could be susceptible to bias, and analyses the results.

However, we should also be aware of other potential issues, such as those pointed out in AI Predictions.