In this day and age, we’re often told that AI can solve all our problems. Just collect vast amounts of data and train a model. It sure sounds great in theory, but reality isn’t that simple.
In the real world of quality inspection, where 80% accuracy isn’t nearly good enough, there must be a better way to determine which applications will truly benefit from these new tools—and which will struggle. Drawing from 30 years of experience using neural networks in various machine vision applications, Ned will share anecdotes and examples to help you predict which kinds of applications are going to be AI wins and which ones are going to give you fits.
Learning Objectives:
- AI: learn how to recognize what it is and what it is not.
- What are the reasons AI-based inspection systems struggle to get to > 99.9% accuracy?
- What can we do to improve the accuracy of these new AI-based tools?
- Could there be value in mixing traditional software inspection approaches ("rule-based systems") with new AI-based tools?
Sponsored by:
![]() |