In today’s episode of #ThisWeekInInnovation we will be discussing bias in #AI. This is a problem that’s affecting many organizations. An AI bias typically means some sort of an irregularity in the data or the algorithm has seeped in and is creating prejudiced outputs. These can be unintentional prejudice created for various demographics like race, color or other variables. There are two forms of biases.

Cognitive Biases - Human biases could seep into machine learning algorithms, either designers unknowingly, introducing them to a model or a training data set, which includes those biases.

Data Bias - Simply the lack of a complete data. It is not be representative and therefore that lack of information can be biased.

We will be talking about how these biases are created, what they are, some of the implications and how organizations could be more aware of these problems and methods to address them.