By asahifile 
What the article is
“Why Gen AI Feels So Threatening to Workers” (pp.54–62) by Erik Hermann of European University Viadrina, Stefano Puntoni of Wharton, and Carey K. Morewedge of Boston University’s Questrom School. It joins the psychology of motivation to the empirical record of generative AI at work.
What it argues
Three psychological needs decide the reaction: competence (being good at something), autonomy (being in control of one’s own work), and relatedness (having meaningful connection at work). When gen AI is experienced as satisfying them, workers treat it as a copilot. When it frustrates them, they resist, withdraw, and occasionally sabotage. The authors’ reading is that most workplaces are currently in the second camp.
The numbers are the argument. A 2025 Kyndryl survey across 25 industries in eight countries found 45% of CEOs believe most employees are resistant or openly hostile to gen AI at work. BCG’s 2025 survey found 85% of leaders and 78% of managers using gen AI regularly, against 51% of workers. In a 2025 survey of 1,600 US knowledge workers, 31% admitted actively working against their company’s AI initiatives — 41% among Gen Z — while 54% said they would use AI tools without approval and 32% use gen AI at work without telling their employer. Only 36% of employees in the BCG study felt properly trained.
The prescription is a five-step framework, AWARE: acknowledge how gen AI may affect the three needs; watch for adaptive and maladaptive coping; align training, mentoring and feedback with those needs; redesign roles around human–AI complementarity; and empower workers through transparency and participation. PwC’s “prompting parties” and “activators”, and Citi’s AI suite across 80 markets and 175,000 employees with more than 2,000 AI champions, are held up as working examples.
The contrast the authors draw between physicians and screenwriters is the most useful part of the piece. Radiologists mostly welcomed a model that flagged urgent cases in 24 seconds rather than the 24.5 minutes a human needs, because demand outstrips supply and the tool relieved a burden rather than reassigning status. Screenwriters read the same technology as an attack on the apprenticeship ladder: if the entry-level room is automated, nobody accumulates the credits that make a showrunner. Competence is not a feeling in that case; it is a career structure, and the tool changes who gets to climb it.
That is where the generational split the authors mention comes from, and it is worth taking seriously rather than treating as resistance to be managed. Experienced workers think younger colleagues misuse gen AI; younger workers think senior ones refuse to change. Both are describing the same thing from different rungs. A framework that asks managers to “acknowledge concerns” without touching the ladder answers the mood and leaves the structure.
Where I push back
The threat is not really about efficiency. It is about the social meaning of competence — the answer to “what kind of person am I?” gets thinner. That is why town halls and training sessions do not fix it. The data also shows that resistance and concealment grow from the same root: oppose it in public, use it in private. That is not employee hypocrisy; it is what happens when an organisation says “adopt this” without redesigning meaning or roles.
AWARE’s weak step is the first one. “Acknowledge the anxiety” becomes a ritual unless the actual allocation changes — headcount, scope, who decides. The 45% figure is a picture of the split between those who deploy the tool and those it is deployed onto.
What it means from Japan
Japanese job boundaries are vague, and work has run on overlapping roles rather than defined jobs, which makes the article’s “redesign roles” step easier in principle. In practice, adding gen AI on top of an undefined job is the fastest way to erase autonomy: nobody can say which part of the judgement is still theirs. The most dangerous pattern is training that is completed on paper and never used.
None of this is an argument against adopting the tools. The BCG gap — 85% of leaders using gen AI, 51% of workers — is a productivity problem before it is a morale problem. The point is that the gap will not close through licences and training hours. It closes when someone senior says, in specifics, which decisions stay human and what happens to the people whose tasks move.
Who should read it
Managers rolling out AI, HR and learning teams, and anyone on the receiving end who feels the tool was done to them rather than with them.
Quotes
“We’re saying we need to hold the keys. Because when companies hold the keys, we get cut out.” — Raphael Bob-Waksberg, creator of BoJack Horseman
“AI is creating uncertainty for all of us, and we can respond to this with fear or curiosity.” — Luis von Ahn, CEO of Duolingo
Source: Harvard Business Review, March–April 2026, pp.54–62 (Erik Hermann, Stefano Puntoni, Carey K. Morewedge)