How AI will Transform Software Testing?Go Back
- Posted by Eric
The ecosystem of software testing is expanding without a pause. Mobile applications built today are communicating with other applications through APIs, they make use of the legacy systems and they are growing in terms of complexity in an evolutionary manner. The point being, how mobile app testers are concerned with this?
According to a recent survey, the artificial intelligence will come to the rescue. The report states that the most important solution to overcome these increasing QA challenges will be the emerging introduction of machine-based intelligence.
But how this will affect us as testers? How will we be using AI to check these growing code suites? What transformation will it bring as it paves the way into our production applications?
Here are five ways in which artificial intelligence will transform software testing.
#1. Our tools are going to change first
Jason Arbon from test.ai is a developer and tester who has experience of working with Google & Microsoft. He has co-authored a book titled How Google Tests Software. When he was asked to comment on the change that will be brought up by AI in software testing, he shared a remarkable yet funny example.
He said that his kids often laugh at him for making the gesture of manually rolling down a car window. He referenced this with the next generation of software testers saying they too will laugh at our practices of selecting, managing and driving systems under test (SUT), for AI will do it a lot faster, better and cheaper.
#2. AI will change our perspective
AI is going to change our perspective software testing and this will transform how software testers work. Jeremias Rosler who has a PhD in CS says that AI’s interaction with the system multiply results that you’d typically have with manual testing. Currently, he is working with on ReTest, an AI-based web app development company that generates test cases for Java Swing applications.
If this isn’t enough to change your perspective, Infosys is offering “AI-led QA”. What does this mean? This means that they are using system data from their current QA systems (defects, source code repo, resolutions, etc.) to help the defects in the software that is to be tested.
Considering this, if AI can help us lessen our work as a software tester, we ought to give him a BFF status.
#3. Determinism will be obsolete
One of the surprising conclusion I’ve reached to is that fact that the problems we solve with AI are not deterministic. If they were, we probably won’t be using artificial intelligence to solve them in the first place. At the same time, one must note that solutions to our problems change over time as our system incorporates new data.
Moshe Milman from Applitools suggests that AI-powered solutions will be in the broad range of possible outcomes. Due to this, the test engineer would be required to run the test multiple times and make sure that statistically, the conclusion is correct. If you’d look closely, this is quite different from our current work processes. It is not only more experimental but through-provoking and more mathematic.
#4. We’ll become masters
One critical question that remains unsolved is what happens when test applications and system under test both use artificial intelligence? Jeremias Rosler recently commented citing the Oracle problem that automation surely knows how to interact with the testing systems but it lacks a great depth when it comes to classifying which is correct behaviour and which isn’t.
Humans are savvy at solving this since they can always approach a product manager, customer or a stakeholder. But how will artificial intelligence do about this? In the upcoming years, software testers will be required a whole together with a different kind of skillset to create and maintain AI-based test suites that test AI-based products.
#5. Test Engineers will be extinct?
I recall a comment from Jason Arbon from test.ai when he said that he couldn’t recall a single activity in the past that couldn’t eventually be done with the help of AI in a better way provided there are enough training materials and data. This sounds like a relief and there is probably a long time between today and when AI takes off.
However, it’s quite comforting to stick on our own importance while we think that AI still can’t do this or that. Yet the best thing would be to acknowledge that whether today or tomorrow, AI is coming!