Automating the Automation using AI

Room 1

Despite the broad research activities in many industries, AI / ML applications do not seem to be present in software test automation. But the use of new technologies such as “artificial intelligence”, “machine learning” and “data science” makes it possible to achieve a significantly higher degree of automation in the overall process of test automation. A research Austrian grant won by Nagarro becomes pioneering work and explores how test automation can be applied with the help of AI / ML. Therefor, an overall concept was developed in order for the test automation framework to add the content below. A few examples of how the remaining manual steps in test automation can be automated:

  • Time-boxed Execution – In doing so, suitable test cases are automatically selected, depending on specific & (current) current risk factors and the desired time window for the tests, and it will not always go through the same test case set.
  • Self-Healing UI Identification – Changes to the user interface automatically adjust relevant test cases and avoid unnecessary errors and therefore time expenditure due to incorrect test cases.
  • Intelligent Test-Run Analysis – The automatic analysis of log files significantly speeds up the evaluation of test results.
  • Defect Reporting – By automatically analyzing the causes of the error, the error correction can be significantly shortened, since with already known error images, no new error ticket is created, but the information is assigned to the already existing error ticket.
  • Change Impact Analysis –There is an automatic selection of relevant test cases based on the code changes made.
  • Fuzz Testing  Due to the automatically generated “random” test cases, a higher test coverage can be achieved and previously unrecognized errors can be found.
  • Automated Test Design – Test cases are generated automatically from user stories and requirement documents.

Key Takeaways

  • AI/Ml can be used to automate the remaining manual tasks from test automation
  • The use of AI/ML from this presentation can be adapted in any type of test automation framework
  • Extensive research is still to be conducted and will change the way we do automation
Automation Track