Philip Lew
XBOSoft (USA)
Metrics for AI in Software testing
With everyone on the AI bandwagon, and many skeptics as well as early adopters, it’s sometimes easy to leave behind basic principles of evaluating if AI in testing is working for you or not. Evaluating the performance of AI in software testing involves looking at various metrics to assess its effectiveness across different stages of the testing process. This includes several metrics in each stage including:
- Test case generation and coverage – Does your AI testing implementation generate test cases that cover what you want? Do the test cases generated cover the requirements you specified?
- Test execution and defect detection – Does your AI implementation detect more defects than you could without it? What about false negatives and positives? Is AI better than humans? What types of priority does AI find?
- Efficiency and scalability – Are you saving time and resources by implementing AI in your test process?
- Quality and reliability – Are your test metrics resulting from AI efforts reliable and consistent?
- Business impact – How much time and money are you putting into your AI testing effort versus what you’re getting in return.
In the end, AI, like any new technology or process, must prove itself worthy in the long run. Without measuring its effectiveness, you may be spending a lot of time and resources without actually improving the quality of your software.
Comprar Tickets

Philip Lew is the CEO and founder of XBOSoft, a firm specializing in software quality consulting and software testing services. With more than two decades of experience spanning engineering, product management, and executive leadership, he helps clients tackle complex software quality challenges across expanding digital platforms. Philip holds a Ph.D. in Computer Science and Engineering from Beihang University, with research focused on software quality and usability, and also earned degrees in operations research and engineering from Cornell University.
A globally recognized thought leader, Philip regularly delivers keynote talks and workshops at international software and management conferences, where he’s known for weaving together technical insight, storytelling, humor, and real-world examples. He currently serves as President of the Pacific Northwest Software Quality Conference (PNSQC), one of the longest-running events in the field.
Throughout his career, Philip has worked with hundreds of organizations to assess and improve their quality processes using measurement, predictive modeling, and systematic improvement frameworks. Outside of work, he’s a passionate cyclist and avid traveler who has explored over 70 countries. His commitment to continuous learning and embracing diverse perspectives fuels both his work in software and his worldview.