Can test automation run without human intervention ?

A common assumption with regard to test automation is that automated test suites can be executed with zero human intervention. After all isn't that what the tool vendors claim their products can do. Theoretically, you should be able to move your human testers to other tasks once they complete automating their tests.

In the real world, automation does not make human testers redundant. Almost all automation test suites, require human intervention in order to remain effective. Consider two simple instances requiring skilled human intervention - analyzing the results of automated test execution and maintaining the automated tests. Also, one must realize that practically getting a complex automation test suite to execute without issues is itself a difficult task.

When the underlying product being tested changes, it is but natural that the automation which tests the product is affected. Even seemingly minor changes to the product, can require fixes to the automation tests. Regular monitoring of changes and the automated tests requires skilled human testers. In the real world, it is common to find that various external factors such as issues with the file system, memory, networking, product dependencies, etc., can also easily disrupt smooth execution of automated tests.

We must also remember the fact that test automation development is very much a software development project by itself and must be treated as such. Like any software that is developed, automated tests are also not bug-free. Regular testing of the automated tests and monitoring of their execution is essential so you can know whether your automated tests are doing what you expect it to be doing. Any changes to the automation tests must follow a similar process akin to a comparable change in the software product, requiring reviews and testing to make sure fixes do not introduce additional defects.