Software testing is a wide and dynamic field, and those that want to remain at the cutting edge of the field should not shy away from technological advances for fear of it impacting the field negatively. This seems to be a perpetual cycle- we saw it a couple of decades ago when the dawn of test automation seemed to threaten the mere existence of manual testing. Yet, manual testing is still with us as a valid and valuable discipline. The arrival of AI, and natural language-based models like ChatGPT in particular, is raising the spectre of a discipline in danger of becoming completely irrelevant once again.
As the Head of Architecture and Consulting at Inspired Testing, I’m tasked with looking to the future to ensure that we continue delivering software testing that keeps pace with technological advancement. While there are risks involved with using AI in software testing, it is merely another tool (just like automation) that we’re adding to the toolkit in our quest to become well-rounded software quality professionals.
Evolution of Testing: From Automation to AI
Thinking back, the arrival of test automation forced a much-needed change in the way we approached testing. It changed the way we test -without making manual testing obsolete- and changed our skillset. Automation allows us to use our time better by doing more risk-based testing, honing our skills, and increasing the speed of providing quality-related information.
If used correctly, AI will have a tremendously positive impact on software testing. Rather than making testers obsolete, it will augment their capabilities and improve the testing process, and ultimately, the quality of the systems under test.
The Intelligence in AI
AI and automation have another thing in common: both require test intelligence. Blindly using any tool to achieve a goal without much thought to the process and end goal merely makes for inefficient processes. The use of AI demands testers that truly know what they’re doing. We refer to it as the I in AI. A tool is only as good as the information you feed it, and this has never been truer than it is with AI in testing.
The intelligent use of AI as a tool demands that a tester knows exactly what they’re trying to achieve and the process to get there. While tools like ChatGPT are smart and becoming smarter every day, it does not have the innate intelligence that a truly good software tester has, and the information it provides should always be checked for accuracy. While AI can provide valuable assistance, using it with caution and maintaining a strong grasp of testing principles is essential. Testers must leverage their expertise to provide the necessary context and ensure that the tool understands the nuances of the system being tested.
It almost goes without saying that the utmost care should be taken with sensitive corporate information, but there have already been multiple instances of sensitive information leaks due to the indiscriminate use of AI tools. At Inspired Testing, we are guided by the ISO27001-certification requirements and follow the guidelines set out by the General Data Protection Regulation (GDPR) and the Protection of Personal Information Act (POPIA). We don’t use customer-specific information when working with AI tools, and I strongly suggest that anyone else follow the same approach.
The AI Assistant: Faster and Better Testers
We fully embrace AI as a valuable tool for enhancing our testing practices and are actively investigating ways to do so safely and efficiently. Language-based models act as testing personal assistants, providing increased test coverage and automation capabilities at an unprecedented speed. It is still just a tool, and its true value to software quality assurance lies in what it allows testers to do: become faster and better testers. By improving test coverage and speed, AI tools will allow testers to undertake more risk-based exploratory testing and undertake more complex testing tasks.
The use of AI in testing is an inevitable step forward- it’s a case of keeping up or getting left behind. The perceived threat of AI will force many testers to shape up and sharpen their testing skills to remain relevant. Now, more than ever, a principle-based approach, rather than a focus on a specific tool or technology, is key to utilising emerging technology to advance software testing as a discipline. All this will result in a net benefit to software quality as a whole.