Are Human Testers Becoming Obselete?
These days, it is easy to get caught up in the hype of AI. Major corporations are investing billions and you can’t interact with anything online without running into AI powered searches, chats, recommendations, etc. You can argue whether AI is good or bad for us, but you can’t deny that it affects all our lives now.
What about software testing? Is AI on the verge of doing it better, faster and more perfectly than a human could? Will AI make regular human testers obsolete? I can’t speak for how things will be in 10 years, but right now my answer is NO.
I recall a similar argument a few years back when automation became mainstream. At the time, one school of thought was that there would no longer be a need for manual testers. And although automation did replace some manual testing, manual testing has remained relevant and integral to the industry and to software quality. One reason is that automation requires different skills and a relatively large up-front investment that makes it impractical for smaller projects and smaller budgets. Another reason is that manual testers are simply better at some testing tasks than automation, such as testing areas undergoing active development, testing visually-heavy areas of the application and testing areas having to do with subjective user experience.
If you replace “automation” and “manual” with “AI” and “human” in the last paragraph you get a very similar argument for the impending obsolescence of humans in the testing industry. And just like manual testing still has its place, humans will continue to be needed, and for all the reasons you would expect. Even well-trained AI can’t match human subjectivity when evaluating parts of a software application that require contextual understanding or understanding user experience issues. Plus, using AI effectively to test a software application is complex and requires skilled staff. Also, professional testers don’t hallucinate.
Don’t get me wrong – the combination of AI with automation is a powerful and useful tool. But I find that with every piece of software I test I notice examples where “AI wouldn’t have caught that”, and it renews the feeling in me that I would never want to release a product without passing it through my “low-tech” manual testers first.
AI content of this blog: 0%