Is it too late to discover the impact AI will have on the testing industry?

🙋 Questions
, , , ,
1

Is it too late to discover the impact AI will have on the testing industry?

I don’t think so. Yet during TestBash Autumn, Carlos Kidman (@elsnoman) sure as heck emphasized the importance of AI in Testing and what that means for our testing roles. He shared the following tips – paraphrased by me and something Ben Dowen (@fullsnacktester) shared on the event chat:

  1. Focus on the deep skill of testing, not just checking. Checking will be replaced with AI.
  2. Familiarise yourself with general GPT-powered tools. Use them to rubber duck yet don’t rely on them to support your testing efforts. Find a domain-specific tool trained on smaller data sets.
  3. Consider what data you are giving to an AI app vendor. And if you don’t trust them, then you’re going to have to build the AI-powered app yourself.

What else would you add to this list?

4 Likes
2

:question: any tips on how to start these discussions with your QA team and wider business?

How do we increase confidence in the ‘checking’ element so we can focus on the more interesting and complicated tests?

2 Likes
3

Welcome to the MOT Club Emma. Hope you enjoy your stay here.

Yep, definitely a lot of distraction with AI powered products going on. People forget that there are many kinds of AI : Generative AI for art and marketing ideation; General Purpose for selfdriving cars et al; and LLM for large data set processing and answering what amounts to technical questions. And frustratingly people don’t realise that these have a lot of quality surface variations. AI has been around for so long, it’s fun watching the point at which AI coders have finally decided to offload the thing to QA coming along almost 15 years after it all started.

I’m more keen to use AI as a rubber-duck type tool, to save me time resource.

3 Likes
4

While AI seems to be a threat to the coding professions one would like to think that the opposite will be the case for testing. With a lot of AI generated code albeit free of “human error” it will still raise the doubt.

As for the automation of test, as a troubleshooter with no love for coding I see it as an opportunity to get that fixed more easily and hope to have more time to focus on the actual testing.

Of course, that might be very optimistic.