I started my QA career as a Games Tester and one of the first skills I picked up was ‘Exploratory Testing’. This method can help you to find bugs which otherwise may be missed when following scripts. Whilst thinking about how important Exploratory Testing is, I also started to wonder if AI could achieve the same results that a human could achieve. AI performing Exploratory Testing, is it even possible? Would you want AI to perform your Exploratory Testing?
What is Exploratory Testing?
Before we can hope to make the jump of getting some AI power house to perform your Exploratory Testing, we need to understand what the method is? Exploratory Testing is a method of someone (QA/Developer etc) exploring an application. They may use their past experience (if any) and guess where the application may break. This methods gives the person the freedom to create their own scenarios, execute and report on them. Notice the key phrases in this paragraph ‘past experience’ and ‘freedom’.
Fitting AI into Exploratory Testing
Firstly, let’s try to define what AI is? The most simple definition that I can express for AI is the following “machines imitating human intelligence”. In other words, a machine that is able to learn, think and act with freedom. An intelligence system that is able to provide facts and opinions. With this said, how does this fit into Exploratory Testing?
For the sake of measurement, let’s use ‘issues found’ for comparision.
Let’s assume you provide access to an application to two different QA’s, one who has knowledge of the application and one who does not. We can assume that the experienced QA will find more issues whilst the new one may not, the new QA will lack domain knowledge of the application. If we replaced the new QA with an AI system, is there a likely hood that the amount of issues found would be identical between the two?
Talking to Developers
There are times when a QA (human tester for clarification) might find an issue that the developers may not think is one. When this happens, the QA needs to ‘convince’ the developers why it is an issue.
For instance, there may be a scenario that causes network connectivity between a backend and frontend system to break. The developers may think that this is not an issue as network connectivity is outside the scope of the application. It would be up to the QA to convince the developers. If we replaced the QA with an AI, would it be able to perform the same task?
AI Performing Exploratory Testing
The big question, is AI able to perform Exploratory Testing?
I would say the short answer is perhaps NO. An AI system (if done well) may be able to perform the same user flows that a QA would for a given application. But the AI would lack the human sympethetic side. It would not always be able to play it’s case when finding bugs. It would lack the ability to ‘convince’.
What about the opposite side? How would humans interact with an AI system? Would a human developer trust an AI system to find bugs?
AI can certainly have a place inside Software Development but would you want to use it? Let me know in the comments section below.
