as a specialist in this field, specifically (test engineering and automation) I’ll say AI shouldn’t be doing this job, but it will and it will probably be better at it.
rationale being: good testing requires deep critique and reflection about product specifications, which are things that product/development teams constantly fail to do properly. there’s a level of abstraction needed to understand what is really needed and expected but it’s not being written down granularly enough, that I think AIs should have a hard time with
at the same time, the average test professional is very incapable, due to how the market is shaped and long-standing neglect of the discipline by institutions. it’s a discipline that is more complex and difficult than the one of development itself, but it’s being done by much much less skilled people
you shot him right in the hill