Avoiding Social Bias in AI Systems - Troy T. Taylor

!mpact
2 min readJan 15, 2021

As we move deeper and deeper towards our reliance on AI and Machine Learning to make decisions on our behalf, it is crucial that these tools are not built with the capacity to be infected with the pitfalls of human bias.

At the onset of their development, AI are like children in the sense that they are born innocent of bias but later taught bias through their environment and training. If their basic education is filled with bias, no matter how unconscious it may be, as these systems develop and become independent of their human creators these biases will remain. This partiality could expand and infect their decision-making processes.

There will come a point when we will not question the AI system’s decisions nor will we be able to intervene in the results. As a consequence, the real-world risk of bias exists. If we are not vigilant and fail to create specific governance policies and procedures, these systems could potentially introduce unwanted biases that will limit the promises of a diverse, equitable, and inclusive world.

Should corporate governance include DE&I in the world of AI and Machine Learning? Isn’t it a question of sustainability? How Machine Learning is influencing Diversity & Inclusion by Pierre Debois is a must read!

--

--

!mpact

!mpact Magazine is a platform where people with a vision can share their ideas and insights.