I came across this interesting perspective about segmenting automation for Software Testing into five areas, based on the activities. It is another way of looking at the various activities involved in automation into different dimensions, and looking at who can do what.
I would strongly opine that the Test Orchestration (you can call the ‘master control’) should be done by humans. This is closely related to the Think/Analyze portion mentioned in the write-up. All said and done, machines have not evolved (not at least yet) based on their experience of the surroundings (like human beings do), and hence they don’t have the ability to think with new perspectives about a situation and take a different route based on their natural intelligence (again, at least not yet!). This is also because humans are the best to judge and decide what to test and which tests to run in a specific instance.
The Act/Service portion mentioned in the write-up should continue to be done by the machines, as they are quick to be done, and do not need intelligence (just an execution of what’s asked to be done).
The Remember/Knowledge portion can be done by machines to the extent of storing the data/information, but information does not directly translate to knowledge. It needs humans to convert information to knowledge. So, here again, human expertise would be required.
In the Testing industry, we are enthusiastically trying to figure out the sweet-spot for the man-machine touch-points, and this write-up provides a different perspective in exploring these dimensions.
Contact me for strategy consulting on Software Testing and deciding when and where humans are required in Testing activities, especially in CI/CD scenarios.