It sort of questioning what QA, QE’s and testers do these days, a lot of companies may be finding out they did not actually know what these groups were doing and as a result their expectations are way off
Some teams on the face of it will get easy wins if they had lower critical skill usage before, those with high critical skill usage its going to be more challenging as less to hand over to machines.
The downside though of flagging repetitive, heavy maintenance or activities with less critical thinking as your main gains is that it can be perceived as just catching up with others, it’s good to do it but it can carry negative past behaviour connotations.
A lot of AI currently favours system of work activities and there are going to be expectations of almost fully autonomous on this front, orchestrating agents level and then autonomous verification build into code generation as standard practice so maybe no need for separate role. This one is going to be a lot about matching expectations and considering what we lose in the transition. I think it’s hard to know what expectation should really be on this front, tool vendors on one hand make it harder as they increase expectations, I think we will get a few QA rightly or wrongly just accepting the tool usage here due to management expectation pressure, hey look industry standard tool cannot argue with that.
Product discovery focused testers will also find expectations challenging, improvement to the system of work changes the risk profile of the product but it does not negate what they do, they are very fast already and the tools in many cases empower them to find more things of value and maybe even sometimes quicker. The primary skill though is critical thinking, analysis and decisions and do you really want to hand over critical things completely to machines, your claude desktop model for example can still keep you at the helm of the critical parts. My feeling is this is not so much a productivity discussion as this role was never a bottleneck but there will be some who feel it is no longer required if their belief in system of work overwhelms product risk consideration.
Token usage targets, we could do a lot of harm on that front letting it run dynamic risk research, hypothesis loops on every PR.
I think a lot will be in the same situation, I am also working in a very pro ai environment so I also get these comparisons with developer progress. So far my skeptical tester hat has been allowed to remain on, building that autonomous orchestra of agents is in cautious progress but I am also pushing hard to maintain the high critical thinking activities and not replacing them with something faster but considerably lesser in value.
I do not have the answers but it is a very valid consideration.