Sovorel reviews a workplace study involving 523 early-career KPMG professionals completing realistic client-analysis tasks. Researchers compared their work with an AI-only baseline and grouped participants as apprentices who performed below it, delegators who roughly matched it, and amplifiers who exceeded it.
The strongest performers did not simply possess better foundational skills. They applied those skills throughout the interaction: framing the problem, specifying evaluation criteria, testing assumptions, requesting alternatives, finding missing evidence and iteratively refining recommendations against domain knowledge.
The result suggests that employers and educators should assess how people direct, evaluate and extend AI output, not only whether they know AI concepts or produce a polished deliverable. Book, certificate-course and channel-support promotions are omitted.
Watch on YouTube
