【Session Theme: Trustworthy AI】
Nowadays, there are many algorithms claim that they can know what we like. The algorithm “COMPAS” are even used to predict if a prisoner would commit again In some states of America. However it comes after that the algorithms may have bias on specific people. For insensible algorithm, how do we define “Non-biased”?
The algorithm “COMPAS” designed by North Pointeic are used to predict if a prisoner would commit again. Despite North Pointeic has pointed that they train the model with previous conviction, age, first commit age, education level etc, the criticism from media never stop.
Suppose North Pointeic is honest, the correctness of algorithm is still need to be verified.
From this topic, I want to introduce how different groups judge the algorithm and share my thought about this event.
About BK Huang
An ordinary programmer.