A coworker was asking about a couple of testing tools, Testim and Applitools. Specifically, he was wondering if either of those had the capabilities to:
Start with Acceptance Criteria or test descriptions and feed them into the tool such that you end up with a PR for tests that cover said AC.
So a flow might look like:
Have AC/Test Description
Feed that text into the test tool
AI tool outputs test code.
AI files a PR for the above code.
Does anyone have experience with those tools, or with that workflow?
Not sure, why anyone would even dream this year that letting an AI tool generate code and put in a PR (pull-request). I mean for that to even work, the tool would have to go through flows of checking that a test even “does” what it says, and then running the test in a suite and not in isolation, a dozen time to ensure it does not leak state. Flows which most of us have not, to date, automated anyway. Sure the tools can generate tests, but are those tests really super valuable yet? I write tooling that generates tests sometimes, but even those are not terribly revealing.
“Might.”
So that’s like a similar question someone asked me about 30 years ago, and I even tried to do some of it at the time, but, it does feel like we are very close now. But nope we still cannot find a machine that will make toast just right, if we did, someone would suddenly start a war, and I’m not seeing any war brewing. My bacon on toast is safe for now.