All fair comments and I appreciate what you’re saying. If you read the article I was careful to say “might be able to”, indicating that there is a possibility these things could happen. The other point worth making is that the list of ideas is based on us inputting information into an AI based application and using it to accelerate the analysis of customer feedback, looking for patterns in date we’ve uploaded, simulate based on uploaded information, analyse complex date sets etc etc. I agree totally that if you ask questions of AI it will be reliant on the existing training data set and any biases introduced by the trainers of the AI. So I wouldn’t ask it questions in that way, I would only use it to make what we do more efficient. You are absolutely spot on with the comment about talking to people involved in the process, could not agree more.

One of the strengths of Six Sigma is that it presents data in such a way that it can be used to make business decisions and to test hypotheses. AI bakes in existing bias and relies on the information already in its training dataset to produce output– this means that it can’t find issues that are not a) Common issues, or b) already visible to the business.

There is some value in pointing out common problems, but you don’t need Six Sigma for that, you just need to talk to people involved in the process.