So I have written about my thoughts on how Large Language Models can help with testing with some examples in the first chapter of my book, but in the spirit of uplifting community I want to call out this great blog post from @bethtesterleeds that I found inspiring when I first started:
What I like about this post is that it shares a concrete example of how specifically Large Language Models can be used to support testing.
Algorithmic versus Heuristic support
Going under the surface of Beth’s post, she is demonstrating how AI really shines when supporting testing activities that are ‘Algorithimic’ in nature. This is concept is something that is shared in Nicolas Carr’s book The Glass Cage. Algorithmic activities are usually well defined, procedural and structured. They will have some form of variability within those confines though, so we can think of it as structured complexity. This differs from heuristic based activities that are less defined and require more implicit and tactic levels of creativity and feedback to solve.
When thinking about how AI can help in testing, focusing on areas which are more algorithmic in nature is where we can get value. That’s what Beth’s article demonstrates. Test data is complex, but it is bound specific syntactical rules. Meaning AI can help us move quickly in those spaces.
So in summary, Beth’s article rocks because it’s fun to try out and has practical applications. But it also demonstrates how being targeted in how AI is used in testing is where the real value is found.