I mean yeah, shit sucks rn and I feel like its definitely possible that companies that are more likely to implement AI are already worse to work for. Psych isn't really my field but I did briefly read the paper and it seems to me like they were pretty thorough. They used multiple companies in different locations and forms of AI, across a few variations of the theme.

Method 1: Three week study, same company & position, no intervention, just tracking correlation between AI use and social variables.

Method 2: Three week study, same company & position, AI usage intervention (as much as possible (test) vs none at all (control))

Method 3: Different companies & position, one-off 30 minute business simulation experiment, AI usage vs no AI.

Method 4: Three week study, same company, different position, AI usage intervention.

The paper is linked from the article, but here it is anyway: https://www.apa.org/pubs/journals/releases/apl-apl0001103.pdf