SOURCE: Kay Sever | August 26, 2026
So far in 2026, I have shared what I have learned about AI capabilities (which change continually), AI components, AI metrics, and the complexities of linking AI to existing IT systems. I have also discussed corporate risks that may not be discussed by AI developers (accuracy, depth of knowledge, resources needed, AI integration timelines, what data to share, how to protect its confidentiality, etc.)
Last month in Part 1 we cracked open the lid on using AI to identify and quantify problems in the workplace. We discussed production problems, how we (humans) identify them and what would be required to enable AI to identify those problems.
This month in Part 2 I expanded on using AI for problem identification and will share my thoughts about AI’s ability to determine corrective action for production problems, knowing that AI MUST HAVE DATA to make recommendations about any problem or option presented to it.
Recognition of a Production Problem
Last month we unwrapped the problem of “unexpected variation”, which could rear its ugly head in countless places in a production value stream. I listed examples of the kinds of data that operators look at to diagnose unacceptable variation or patterns outside of “normal” or “best” parameters. One paragraph from last month’s article is worth revisiting here:
Unexpected variation is a common indicator of a possible problem… process variation that exceeds defined min/max limits OR variation from a standard (construction/design specifications, cost, volume, input quality, output quality, density, temperature or other process characteristics). If people are watching a process perform real-time and are expecting a certain consistency or result but do not see the expected result, they know there is a problem somewhere based on what is defined as acceptable/normal operations.
As stated in last month’s example “The 10-Minute Million Dollar Problem”, knowledge of both duration and frequency of a recurring problem over time were required to arm management with the information they needed to prioritize that problem and take corrective action to fix it. In that example, millions of dollars were lost annually for years until management became aware of the size of the loss and implemented a fix for the problem.
Could AI be given the data and information that reveal production problems? The answer may be YES, but a YES answer would require lots of process documentation and AI interface work. The bigger issue is that all problems are not equal! By that I mean some problems occur within a single productive unit (a haul truck, for example). Others occur within a series of productive units connected together (a crusher line or flotation circuit, for example). Still others may occur downstream in another plant but the root cause happened upstream. Still others could be caused by a supplier input that was out of spec, either where the problem occurred or in an upstream plant.
AI’s Ability to Recognize a Problem
For AI to be able to diagnose all of the above types of problems, the entire value stream (each productive unit at a mine/plant) and supplier inputs would have to be mapped with product/process characteristics and standards for those characteristics. Further, things that make a mining site unique could cause AI to miss real problems at one site or mistakenly report problems that do not exist at another site.
Examples of characteristics that would create unique metrics for problem identification and resolution for each site include:
- Variations in production equipment and conditions between operations.
- Variations in geology/ore type, elevation and environmental restrictions.
- Customized equipment for a specific application (i.e., agglomeration, etc.)
As a result of these variations between sites, an AI agent designed to identify mining production problems could have different criteria for problems at each location. For this reason I believe it would not be possible to take a “plug-and-play” approach to installing an AI agent designed to find mine production problems across multiple sites without significant customization work for each location on the AI agent side.
AI’s Ability to Recommend Corrective Action to Solve a Problem
It’s one thing to identify a recurring problem and another thing to make a recurring problem stop happening. To stop a recurring problem from happening again, management needs process knowledge specific to 1) the root cause and 2) the “gemba” (the physical place the loss is occurring in the value stream). The logistics/environment in the area where the problem is occurring can have a significant impact on the solution chosen.
Engineers and operators would know the physical layout/restrictions/limitations of the affected area and could evaluate possible solutions with those limitations in mind. If AI had the capability to recommend a solution to a problem, it might choose a solution incongruent with the physical conditions in the problem area. If the AI agent was programmed to reduce costs, it might choose the cheapest solution which would either not work for some processes or physical locations OR would cause new problems for production.
It is my opinion that, in a mining environment where each site is different (dimensions, material type, equipment configuration, environmental restrictions, physical space within and between plants, etc.), there would be more issues with AI recommending solutions to recurring problems than if AI revealed problems for management to prioritize and solve.
Kay Sever is an Expert on Achieving “Best Possible” Results. Kay helps executive and management teams tap their hidden profit potential and reach their optimization goals. Kay has developed a LIVESTREAM management training/coaching system for Optimization Management called MiningOpportunity – NO TRAVEL REQUIRED. See MiningOpportunity.com for her contact information and training information.Optimization/AI Integration – AI Scope: An Overview.
To comment on this story or for additional details click on related button above.
- About Us
Kay has worked side by side with corporate and production sites in a management/leadership/consulting role for 35+ years. She helps management teams improve performance, profit, culture and change, but does it in a way that connects people and the corporate culture to their hidden potential. Kay helps companies move “beyond improvement” to a state of “sustained optimization”. With her guidance and the MiningOpportunity system, management teams can measure the losses caused by weaknesses in their current culture, shift to a Loss Reduction Culture to reduce the losses, and “manage” the gains from the new culture as a second income stream.
