Biased De-Biasing

In this article from McKinsey, Controlling machine-learning algorithms and their biases, they make the point that while machine-learning is good to avoid human bias in decision making, the algorithms themselves are not free of bias. Let me say it in another way:

“Garbage in – Garbage out!”

McKinsey make the point that the designers of an algorithm are human and that they will come with their own personal bias. For example, they will choose sources of data and/or choose not to use certain data. One can’t get a feed from every data source in the world at once (yet!).

They go on to point out that bias within the data can creep in. An obvious one is spikes in loan defaults following a natural catastrophe. If your organisation does not have a method for dealing with the spike, the algorithm may develop a bias across the portfolio or a geographic region.

The article reminds me of another saying:

“There is nothing new under the sun.”

While every innovation, every piece of previously unheard-of technology is of course “new”, nothing changes the fundamentals of the world we live in. The same basic human needs, from food and shelter to community and love, continue to exist.

We are a world of people making decisions to act or not act. So remember:

“Your decisions define you.”

My favourite saying!