How optimistic are you about the potential of AI to solve existing challenges in software testing?

🙋 Questions
,
1

How optimistic are you about the potential of AI to solve existing challenges in software testing?

Feel free to add comments for more context and insight.

  • Not optimistic
  • Somewhat optimistic
  • Very optimistic
  • Extremely optimistic
0 voters
2 Likes
2

Personally I’m somewhat sceptical of the idea of AI taking full control of things. I’m more than happy to say that AI makes for a good augment software testing (i.e. helping to optimise automation scripts, spotting areas that may have been missed, etc.) but I don’t see it as the solution to all problems.

4 Likes
3

I voted not optimistic at this point in time. I have trouble trusting the outputs of the popular LLMs where they are trained on the entirety of the internet. Garbage in > Garbage out comes to mind and when the trained dataset is the garbage that is the internet then I can’t see myself trusting these LLMs without having to double-check it myself, in which case I might as well have skipped asking the AI in the first place.

I would be more optimistic when there are company hosted internal LLMs where we can control the training set and can more easily validate the outputs.

1 Like
4

At least for the forseable future, AI is most usefull for kinds of fuzz testing, finding combinatorial errors efficiently in crazy paths and random explorations to yield insights where even a child with loads of patience could break it type testing. Testing which we frankly dont give enough time to anymore. And api security is just such a target, and credible tools are already in play in the sector, just without the AI label anymore.

5

I think AI will prove to be useful in the ways that it is already showing: boiling the ocean.
With fuzz security testing and best practice gating at the pull request level, it is already showing benefits. The challenge that will remain will be separating the noise from the signal. Thats where Quality Analysts will earn their pay: refining those gates and ocean boiling AI tests to raise issues that are valuable and to squelch those that arent.

6

AI has great potential. But it required an adequate approach and the right tuning. And no, it can’t replace QA Engineers like it can’t replace real engineers in other areas.

1 Like