At the end of last year, I had the opportunity to participate in two events that gave a lot of inspiration; EuroSTAR in Hague and Sogeti Testing Community Event. The result has been a lot of reflection during the holiday. I realized there were a lot of things on automation and AI that I haven’t formulated before. Therefore, I’m planning to write down some thoughts around automation and AI. It would be great if you comment with your own thoughts on the subject… 😊
It’s an exciting time now in the testing field because a lot is happening both with the systems and with the tools available for testing. One of these is the possibility to automate test case design. One approach is to generate test cases from a model of the system. If you complement it with the automation of the test cases, you can save a lot of time, but there are some things we need to think about and find a solution to.
A model is an abstraction of the system and does not include all information about the behavior of the system. It also doesn’t include the self-evident requirements that weren’t formulated and therefore is absent from the model. All necessary test cases can’t be generated from the model, because the model doesn’t include all the information. This means a tester needs to design the “missing” test cases because those test cases are most likely to discover bugs (missing information is often the source of bugs). A tester builds his/her knowledge while working on designing test cases and testing the system. Often you start with the simpler, more apparent ones and take the more complex ones later. If we automate the designing and execution of the simpler test cases, how will the tester build his/her knowledge enough to be able to design and test the more complex cases?
Humans tend to trust technology more than themselves. This has led to some tragic accidents there for instance pilots have trusted the instruments more than their own eyes. If we automate the design of the test cases, how can we mitigate the risk that the tester relies too much on the generated test cases and doesn’t trust his/her own judgment?
The issues above are possible to solve, but we need to start discussing them to find the best solutions. What do you think?
About Eva Holmquist
Eva Holmquist has more than twenty-eight years of professional IT experience, working as a programmer, project manager and at every level of the testing hierarchy from a tester through test manager. She has also worked with test process improvements and in test education as a teacher and with the development of courses including a Swedish ISTQB Foundation certification course. Author of the book ”Praktisk mjukvarutestning” (Software Testing in Practice) as well as science fiction and fantasy novels. Eva works as a Senior Test Specialist at Sogeti helping clients improve their testing practices using her broad experience in system development, process improvements, and education. She is a frequent speaker and has during the last year held presentations about agile testing, DevOps and quality assurance, cognitive quality assurance and bias in artificial intelligence.
More on Eva Holmquist.