Whether you are steering your navigation system in your car by voice recognition or telling your smart speaker at home which song to play, Artificial Intelligence (AI) is behind it all. Machine learning and AI systems have conquered the tech world and drastically influenced our lives. It’s the logical next step that this technology also revolutionises software testing. However, what is already possible with AI in software testing today – also referred to as AI testing – and how will it change this sector in the future?
The consultancy Capgemini states in its World Quality Report 2016/2017 that the “introduction of machine-based intelligence (…) will be the next big wave of change after the introduction of risk-based strategies and test automation technologies.” Though test automation is known as an efficient method, especially for regression tests and companies using a continuous integration approach, many firms do not use it as much as they do other testing methods. According to the Capgemini World Quality Report 2017/2018, the average level of automation for test activities is only around 16%. If AI testing is the next step, does that mean it will still take years until companies use it? Let us have a look at what is already possible in the field of AI testing.
Generally, several use cases for AI Testing are possible, such as a focus on test management and the automated creation of test cases, generating test code or creating and running tests automatically on an application without writing any code. Currently, there are already solutions on the market, which automatically generate tests and execute them on user journeys. Furthermore, some solutions use existing data in a company’s QA system, like defects and test cases, to identify problem areas in the product.
So, what might be possible in the future of AI testing? Humans could have less work with implementing, executing and analyzing test results, but they will be the ones approving and deciding what to do with them. In this manner, AI is able to help us test faster and more efficiently.
When it comes to user experience, human feedback, for example from crowd testers, there is still an urgent need to get a comprehensive overview on how the target group thinks and feels about a product. The reaction of the real user to a product, which is also emotional and not always lead by logical principles, will most likely not be simulated by an AI – or at least not in the near future. Studies in the field of Affective Computing aim to make computers emotionally intelligent. Maybe an AI as sophisticated as Watson by IBM could estimate how fast this field will progress.