top of page

Bias in AI

Bias can creep its way into AI in a multitude of ways. The functioning of AI is based on the foundation of training data - subject to human prejudice, error and fallacies, whether personal or collective. More than thirty years since a British medical school was found guilty of racial profiling by the UK Commission for Racial Equality, bias in AI continues to exist.

 

The two main sources of bias currently are - flawed data sampling and biased human decisions even if variables like race, gender and sexual orientation are eliminated.

Diversity
Wavy Abstract Background

How to deal with bias in AI

The Rule of Six

​

  • Business leaders need to stay relevant with their research.

  • Establish of responsible processes to mitigate bias.

  • Engage in fact-based conversations about human bias.

  • Devise methods in which humans and machines can work together to mitigate human bias.

  • Invest in bias research.

  • Invest in diversifying the AI field.

​

bottom of page