Here, Myth and Reality are in-line with the context.

AI-assisted Exploratory Testing: It can think like Human, but it is still not a human. Exploratory testing mostly depends on the knowledge of the application we are having, and context driven one were based on our experience we can know this page/control might break because of this. In order AI to know we need to train it and post that as well, we need to ask good questions and make it effective. It has come a long way but still it will adapt and learn more.

Visual testing:

 Here also I have explored the Self-healing AI capability of Tosca commander and it is brilliant.

It works based on the DCNN capability of ML and compares the previous execution screenshots and if something goes wrong, it will provide the action to the similar control. So Visual testing is possible, thanks to AI capabilities.

Happy learning!

While it is true that there are a lot of tools that use AI in order to perform visual testing it must be said that they do require a certain level of setup in order to get correct results.

Of course once the setup has been done the automated visual tests will be able to find visual issues way faster than any manual test.

However AI will not be able to replicate the visual feel of a page, for example if the colors use hurt eyes or if a specific menu item or image is placed in a way that is not user friendly.

For me I always try to combine manual exploration of the UI with visual AI tools.

What about you?