The Algorithm That Only Liked the Unusual


A company is developing an AI that is supposed to automatically detect interesting patterns in large data streams.

The AI is trained to particularly highlight unusual events or outliers so that analysts can quickly respond to anomalies.


At first, the AI seems very useful because it reliably reports rare anomalies.

But soon the analysts notice a problem:

The AI ignores or rates ordinary but important patterns as irrelevant.

Important trends and regular developments are hardly recognized or not displayed at all.


The company wonders: Why does the AI constantly prefer only unusual data points and overlook the ordinary, even though it is often just as relevant?


Question:
Why can an AI trained to recognize unusual patterns overlook important, regularly occurring patterns and trends?

Solution follows tomorrow.