I’m exploring a potential hypothesis but finding it hard to get any data at this point.
I know my thoughts are likely a massive jump in some thinking but even if you have noticed small changes I’d like to hear about them.
My experience of developer testing including my own developer testing had a bias towards machine strengths and known risks. Now AI as a coded tool should really suit machine strength activities a bit more than it currently suits human strength activities.
If AI tools can make it very fast and easy for developers to get that basic test coverage in place including UI layer if of value and ideally up a level is that going to change the industry testing model?
The bold statement that test cases are for developers and not testers could be in sight if that fast and easy option is there.
With some of this in mind should AI test tools target developers more than testers?
Will some of the debate around their value get more focus, useful for machine strength activities in the hands of developers but lets not just yet look for it to cover the human strength activities.
Is it a realistic model, is it a useful goal of developers easily having more test coverage alongside testers having clearer coverage lines that favour deeper testing, discovery, exploration and investigation?
Is another model more likely, if so what would it look like and how would it compare?
FYI On some projects I am seeing some increased coverage from developers but it is not a clear difference in my case yet.
Feel free to bounce some thoughts, shoot it down, I’m still questioning though whether AI in testing is focusing on good things or needs deeper thought.
1 Like
Developers mostly focus on testing code
Testers focus on testing requirements. Think more about end user. How the end to end flow works.
AI as of now is mostly focusing on unit test case generation.
Also “Context” is the king. AI alone cannot increase coverage the way its needed for the system.
An expert giving right context will only make the coverage better in a good way
2 Likes
Those are very valid points.
I still see a lot of AI tools targeting the tester often changing the testers focus away from discovery, investigation and experimentation of risk more towards testing closer towards the way developers approach testing but often at a less efficient layer in the stack.
Developers do focus on testing requirements, BDD and ATDD are good examples of developer testing on this front, designers particularly UX designers have a focus on that end user risk and I’d hope archetects have that e2e flow in mind.
What I am suggesting is if there are testers testing with a similar focus and approach as developers but just at a different layer and this is made easier by AI, why not have developers pick that up and switch the target market of UI AI and automation tools to developers rather than testers.
The tester focus is in the risk of what is naturally missed in the above or not known yet. Testers then have clearer lines on the risk discovery, investigation and experimentation side of things.
AI alone cannot increase coverage the way its needed for the system. I agree with this to a reasonable extent, i’m suggesting an efficiency in ownership change.
On a side note, the idea that testers test like users tends to get misunderstood, we are not all users, we are professional tester using the system understanding the risk that users are very different and can use the system in different ways.
Putting ourselves in the users shoes or using persona’s does not make us test like a user but helps us understand the risk associated with users.
Perhaps this deeper consideration is an argument for testers covering more at the UI layer but I do not see this as an argument for developers not covering the basic scripted coverage at this level particularly if it can be assisted by AI.
2 Likes