Book Review: Software Testing With Generative AI

Target Audience

The recommended audience, according to the Manning site:

About the reader: For developers, testers and quality engineers.

While this is accurate, I’d expand on it a bit. The book seems particularly valuable for junior to intermediate-level practitioners who aren’t yet completely comfortable with LLMs and prompt engineering. Those already well-versed in Generative AI might find some sections less beneficial.

The book assumes no prior AI knowledge - concepts are explained from the ground up, which most readers will appreciate. However, you should have some basic software development or testing experience to get the most value from it. While Winteringham does explain testing concepts throughout, I feel that having that foundation will help you better equipped to apply insights shared in the book.

I think a reader should also do a bit more reading up on prompt engineering, even though this topic is covered in this book (this would also help you to derive further value from what Winteringham shares).

What I Thought About the Book

I was excited to dive into Mark Winteringham’s latest book ‘Software Testing with Generative AI’ (also sometimes referred to as ‘AI Assisted Testing’ online). As one of the first books on the market about testing with AI, it couldn’t have come at a better time given how prevalent this technology has become in our workplace.

I had high expectations going in, having already benefited from Winteringham’s other book “Testing Web APIs” and his “Automation in Testing” training (alongside Richard Bradshaw). I’m happy to say this book didn’t disappoint. This book helps you get from 0 to 1 (and then some) when it comes to software testing with Generative AI.

Figuratively, Winteringham is holding your hand through this entire journey. For people who are curious about these technologies but not 100% in how they can incorporate them into their work, with this book they have Winteringham’s expertise in their backpack (or their laptop if that’s your cup of tea).

What really sets this book apart is Winteringham’s practical approach. He doesn’t just tell you what to do - he shows you, repeatedly, with concrete examples of how to use Generative AI in your testing work. Some of his ideas were genuinely impressive, and I found myself thinking “wow, I hadn’t considered that approach!” more than once. Throughout the book, you’ll find activities that let you apply what you’ve learned, making the concepts stick.

I particularly appreciated Winteringham’s balanced perspective. He looks at LLMs and Generative AI with a critical eye, clearly explaining not just the benefits but also potential pitfalls, including how to handle the risk of hallucinations. While he does reference some useful resources, the relatively new nature of this field means the external literature is still somewhat limited.

This book is focussed on using ChatGPT as a tool, but Winteringham acknowledges that the concepts he share in this book can be used on its competitiors.

I accessed the book through MEAP (Manning’s Early Access Program) and waited until it was complete before reading it. I read about 80% as a physical copy and the rest using Manning’s Livebook feature. While Livebook provided a better experience for code samples, I personally preferred the physical copy overall. I’m glad to have access to both versions.

All in all, I’m glad I bought it. I even spotted my husband (a developer, so he is also part of the target audience) reading some pages multiples times when I wasn’t reading it myself. So you could say he is also glad I bought it!