The Problem with AI Bias

You can’t solve the problem with the current structure

If a tree falls in the forest and there is no one around, does it make a sound? Scientists would say “Yes” because the sound waves are created. Sociologists would say “No” because no one was around to hear it. The psychologist would say that reality can only be defined by humans and therefore conclude that reality is dependent on the presence of humans to hear the tree.

What does any of this have to do with AI?

The presence or absence of data generated by humans or the interaction of humans with their environment is the foundation that underpins artificial intelligence. And data scientists are discovering that the data is incomplete and sometimes misleading. The data sets used to train the computer programs are often the source of bias.

But I believe there is a more pernicious problem that will be created if we fail to address the socio-economic factors that underpin data sets.

As a scientist, I learned that data is EVERYTHING. And scientists go to great lengths to ensure the integrity of any data set that is used to validate a hypothesis. The way that scientists go about this is by designing an experiment in which variables are defined and controlled. Scientists do their best to eliminate “wild cards,” that is unknown factors that may impact…