Testing AI-Infused Applications: The Good, The Bad, And The Ugly

For the next five years, you will be infusing your applications with artificial intelligence (AI) capabilities, giving them speech, hearing, sight, and decision-making capabilities. You might be building some of those capabilities yourself by developing machine learning or deep learning models; others you will be integrating in. They all have one thing in common: They learn. How do you test the learning components in your AI-infused apps for bias, for drift, for shift, or simply for not achieving business goals? How do you test those AI components, their integration in your standard application code, or the whole of the application end to end? What tools, technologies, skills, and practices do you need?

Key takeaways:

• This is not about using AI in testing but about testing “intelligent” applications. What do you need to change in your testing strategy and capabilities?
• What are the skills, roles, and technologies to test AI-infused applications? Does testing change? How?
• What should you prepare for? What’s possible, what’s easy, and what’s hard?

Target audience level: all levels

Vendors mentioned: Amazon, Clarabridge, CognitiveScale, Cognizant, Cyara, Google, IBM, IBM GBS, Micro Focus, Microsoft, Nuance, Parasoft, Qualitest, Salesforce, Sisense, and Tricentis.