Editorial Team
Chicago, 16 September 2026. Dr. Jay Lee of the A. James Clark School of Engineering at the University of Maryland delivered the keynote at the first IMTS Industrial AI Conference today, and organised his argument around three barriers that manufacturers hit when they try to move AI out of the pilot stage.
The three barriers
- Identifying usable data. Not whether the plant has data, which it almost always does, but whether the data it has can answer the question being asked of it.
- Selecting the right AI tools. A choice usually made before the problem has been stated precisely enough to make the choice obvious.
- Deciding where intelligence runs. Edge or cloud, a decision with consequences for latency, cost, data governance and who can change the model.
The framing is deliberately unglamorous. None of the three is a modelling problem, and all three are decided before any model is trained.
The case study
A featured session showed how manufacturers can apply low-cost edge AI devices to monitor CNC machines using a stream-of-quality methodology, which tracks quality-relevant signals along the process chain rather than inspecting only at the end of it. It is a method that suits job shops precisely because it does not require a data-lake programme to start.
"We created the IMTS Industrial AI Conference to help manufacturers move beyond experimentation and understand where AI can deliver measurable business value." — Ryan Kelly, Vice President of Technology, AMT
The conference was developed with Lee and the University of Maryland's Center for Industrial AI, and complements the Industrial AI Arena on the show floor, where 32 vendors are exhibiting. IMTS closes on 19 September.
Get the next one in your inbox
Every appointment, deployment and move in industrial AI. One email a week, free.
