Hype Cycle Peak for AI/ML – but what about the basics?
When you start-up a new tech company – one of the great side benefits is the need to deeply understand the market challenges and ensure you are building a solution that meets customer needs and drives business value. As our team has been reading, researching, and analyzing, we are amazed at the amount of hype in place for the use of AI/ML technologies applied to failure prediction of industrial equipment. The reality is when you exclude Original Equipment Manufacturers who are motivated by large parts and services agreements, there are limited industrial use cases at scale with proven benefits.
One challenge with AI/ML technologies is unless you have a clean, golden data set to train the models and/or large populations of similar equipment in the same operating context, the models can prove to be difficult to build and implement with accuracy. Having said that, we have applied AI/ML in certain use cases such as similarity based modeling, asset tag mapping, and long text mining so we are very supportive of the technology but believe the current hype has gone a bit far.
Do not get distracted by the hype around the technology or mathematical approach– there are already proven engineering approaches available which mitigate industrial equipment risk through analytics. Let’s also remember that technology is only one dimension of a solution that needs to be driven by people and processes.This brief by Bain & Company is a great reminder to stay focused on the problem you are seeking to solve, be practical, tailor to the reality of your plant, engage the resources that know the equipment, and retain your independence!