Episode 1: AI in Quality Management #QualityMatters2.0
AI in Quality Management: Useful Tools and Human Judgment
AI in quality management has real potential to save time. Helping users navigate software, reviewing information, and preparing summaries can reduce the effort involved in everyday work.
But usefulness depends on what you ask it to do. Explaining where to click in an application is different from deciding whether an inspection process meets its requirements. At TQA, we see that distinction as essential to choosing where AI belongs.
Start With a Specific, Useful Task
Adding AI to software should solve a problem people actually have. An assistant that looks impressive but gives unhelpful answers only creates another obstacle.
One application we’re considering is a help bot for our quality management software. It could answer navigation questions and walk users through tasks while our team is helping another customer.
That gives the assistant a focused purpose. Helping someone upload a record is a manageable starting point. Giving advice about a quality problem requires a deeper understanding of the business, its procedures, and its circumstances.
Human Verification Still Matters
An AI-generated report might identify patterns in nonconformances or corrective actions. That could give a quality manager useful information to investigate.
The next question is how the report reached its conclusions. Which records did it use? What calculations did it perform? Can you explain the result to an auditor?
Kyle has walked auditors through spreadsheet calculations to demonstrate how he arrived at reported numbers. An AI-generated explanation doesn’t remove that need for human verification. Someone still needs to check whether the explanation matches the underlying work.
Adding AI can also introduce another layer to review. You may need to check both the original calculations and the AI’s interpretation of them.
Power BI Reports Help You Explore the Records
Our app’s Power BI reports let users filter information and drill into the records behind their metrics. Supplier performance is one example: you can explore the data, select useful filters, and return to those filters when reviewing a KPI.
You can also export the report to a PDF for review or an audit.
Power BI reports still need checking. We continue to review calculations and investigate unexpected results. Their value is the ability to explore the information and explain how a report works.
AI summaries may offer another useful view, but they don’t replace that connection to the records.
Auditing Requires Understanding the Work
A document review can tell you whether certain information exists. Auditing an inspection process involves observing the work, asking questions, and following details that deserve closer attention.
Experienced auditors bring knowledge of how different organizations solve similar problems. They can recognize when an explanation looks reasonable on paper but needs further investigation.
We don’t see AI replacing that judgment. Checking information is one part of an audit; understanding what happens in practice requires more.
Keep Advice Connected to Your Business
A supplier approval process that works for one customer may differ from another customer’s chosen approach. An assistant must understand that context before offering useful advice.
Customer information also needs protection. We would need to address both concerns before introducing an assistant with access to customer-specific records.
Our approach to AI in quality management is to start with focused assistance and preserve human verification. AI can help people work faster, but quality professionals still need to understand the evidence, question the results, and make informed decisions.
Podcast: Play in new window | Download