Managers Under the Gun: The Human Cost Lurking Behind the Glitz of AI

Managers are more anxious about AI than the people they manage. There's a 45 point gap between how executives believe their staff are responding to AI and what employees actually report. Employee engagement has collapsed from 88% to 64% in a single year while burnout has stayed stubbornly at 83%. And 56% of workers say their employer never consulted them on how AI would be deployed in their roles.

The AI wave, a revolution in technology and economics, could fast become a leadership crisis. The people best positioned to temper it, middle managers, team leaders, people managers at every level, are the least supported people in the organization.

A team was recently asked to pilot an AI resource, intended to facilitate decision-making and subsequent work streams, record keeping, and comms. The team agreed the content was useful. But then someone said: "I’m so sick of AI."

Consensus was clear, but not because they were resistant to technology. In the subsequent discussion, it emerged that staff are being asked to test AI tools from every direction, HQ, global teams, finance, HR. They were exhausted by the onslaught, the lack of clarity about rollout and coordination, the volume.

They were destabilized by the complete absence of any conversation about what the AI mandate was costing them, the company, or the future of their careers. In the immediate, AI integration was another line of work on top of their regular load. In the medium and long term, there was no strategy or plan. All of that was hidden behind the glitz of new technology.

Because of that comment, made by an employee who felt safe enough to say it, the team could reconsider the ways that AI was entering their workforce.

Subsequently, planning sessions, continued focus on an environment safe enough for staff to say when something wasn’t working, and data to support careful and intelligent adoption of AI, among other elements became consistent in their plan to integrate AI with human intelligence.

That moment is not an isolated incident. It’s playing out in organizations across every sector in 2026. The AI adoption curve is steep. The human cost is real and largely unaddressed.

The people being asked to manage that cost, without training, without honest information from above, often while handling their own concerns and pre-existing workload, are middle managers.

Managers are more anxious about AI reshaping their roles than frontline workers (Mercer Global Talent Trends 2025)

76% vs 31% Executives who believe employees are enthusiastic about AI vs employees who agree (McKinsey Superagency in the Workplace 2025)

88% to 64% Employee engagement drop in a single year while burnout held at 83% (DHR Global Workforce Trends 2026)

56% of workers say their employer never consulted them on how AI would be deployed in their role (JFF National Workforce Survey 2026)

88% of heavy AI users report increased burnout (Harvard Business Review 2026)

95% of companies using AI have seen no measurable return on investment (McKinsey State of AI 2025)

Only 1 in 10 employees globally feels comfortable using AI at work (ADP People at Work 2026, 39,000 workers)

The gap between 76% and 31% is not a communication problem. It is what performative certainty looks like, and may well exacerbate imposter syndrome, itself destructive. That is, managers projecting confidence they do not feel. Teams see through it, eroding trust and confidence. These are not low stakes because AI threatens human beings’ jobs.

Most AI coverage focuses on employee anxiety: the fear of job loss, the resistance to new tools, the burnout from constant change. What Mercer identified is the layer above that: the managers being asked to project confidence and guide their teams are more anxious than the people they are guiding.

Teams are sophisticated. They feel the inauthenticity of false reassurance even when they cannot name it. The McKinsey gap, 76% of executives believing employees are enthusiastic, 31% of employees agreeing, is not a data anomaly. It is what a 45-point trust deficit looks like in survey form.

DHR Global's 2026 Workforce Trends Report quantifies the consequence. Employee engagement dropped from 88% to 64% in a single year. Burnout held at 83%. The gap between those two numbers is a trust gap. Disengaged and still burned out. That is what an organization looks like when the people expected to build trust are themselves unsupported.

The AI mandate is real and it is not going away. The question for managers is not whether to lead their teams through it but how to do so without the casualties that poor AI leadership produces, ex: talent loss, team fracture, toxic competition, etc.

The research is consistent on this point: Organizations that manage the AI transition well through 2028 will not be the ones that adopted fastest. They will be the ones whose managers led honestly, paced the mandate deliberately, and stayed invested in the humans doing the adopting.

That requires something most organizations are not providing, frameworks for the specific conversations AI creates, scripts for the moments managers are unsure, language to push back upward when the mandate is moving faster than the team can sustain.

When AI pressure lands on a team without adequate management, the research identifies a predictable sequence.

Fear activates scarcity mindset → Scarcity mindset activates territorial behavior → Territorial behavior in a workplace looks exactly like the toxic dynamics that cost organizations their best people.

Further, managers and their teams can expect to see the following:

→ Credit theft accelerates. Contributions are guarded more aggressively when people feel replaceable.

→ Information hoarding increases. Knowledge becomes currency when jobs feel threatened.

→ The early adopter / resister divide becomes a team fracture line.

→ Managers under pressure may use AI adoption as a performance proxy, creating unfair and demoralizing comparisons.

→ Psychological safety collapses. People stop raising concerns, stop admitting struggles, stop contributing honestly.

Spring Health's 2026 survey of more than 1,500 workers found that 24% reported worsened mental health due to AI-related information overload, and 23% reported a reduced sense of control over their future. These are not marginal findings. They are the early indicators of a retention crisis building underneath the AI productivity narrative.

The team that tells its manager it is sick of AI is a team that still trusts its manager enough to tell the truth. The team that goes silent and starts building exit plans is the one that has already lost that trust.

These are not technology recommendations. They are people management principles for the specific challenge of leading teams through AI disruption.

1. Subtract before you add.

For every AI tool introduced, identify something removed from the team's workload. AI adoption fails when it adds coordination without removing tasks. Forty-three percent of employees cite excessive workload as their primary burnout driver. AI that adds review layers rather than removing them accelerates that problem.

2. Name concerns out loud.

Mercer's finding that managers are more anxious than their teams is precisely the finding that makes honest conversation so difficult and so necessary. A manager who can say, 'I know this is a lot, and I have questions too' does more for psychological safety than any AI training program.

3. Close the 45-point gap.

The McKinsey gap between executive belief and employee experience is a communication failure with measurable consequences. Regular, honest, two-way conversations about AI, not mandates, not briefings, but genuine exchanges, are the intervention that closes it.

4. Protect the resisters.

AI resistance is almost never laziness. It is anxiety about relevance, cognitive overload, or a legitimate question about whether the tool is fit for purpose. The people who resist are frequently the most honest sources of feedback about whether the adoption is actually working. Protect them.

5. Separate AI adoption from performance review.

Only 45% of organizations have a formal AI policy. Evaluating employees on AI tool usage before adequate time, training, and standards exist is not fair management. And teams know it.

6. Push back upward when necessary.

The manager's role includes protecting the team from impulsive decisions above. When the mandate is moving faster than the team can sustain, saying so, with data, with specifics, with a proposed alternative, is not insubordination. It is leadership.

7. Build psychological safety specifically for AI uncertainty.

Amy Edmondson's research is unambiguous: psychological safety is the single greatest predictor of high-performing teams, and it matters more during periods of disruption. Your team needs to be able to say: I don't know how to use this. I think this output is wrong. I need more time. None of those statements should be professionally dangerous.

Wild at Work has built a clear, six-part guide specifically for this moment, available immediately. It's geared for manager action, and can support team members and senior leadership to assist their managers.

The AI Workplace Field Guide is built on 2025 and 2026 research from Mercer, McKinsey, Harvard Business Review, Gallup, ADP, JFF, DHR Global, Spring Health, and the Wild at Work diagnostic data.

Contents include:

Part I: The Landscape. Data on What is Happening Now.

Part II: The Human Cost.

Part III: The Managers Playbook. Eight Key Points to Structure AI Adoption.

Part IV: Leading Transparently and Authentically.

Part V: The AI Manager Quick Reference Guide/Checklist with Scripts for every conversation.

Part VI: Measuring uptake: KPIs for AI Adoption.

Mercer (2025). Global Talent Trends Report.

McKinsey & Company (2025). Superagency in the Workplace.

DHR Global (2026). Workforce Trends Report.

Harvard Business Review (2026). AI burnout and heavy user research.

McKinsey & Company (2025). State of AI Report.

Gallup (2026). State of the Global Workplace.

ADP Research Institute (2026). People at Work: A Global Workforce View. N=39,000.

JFF, Jobs for the Future (2026). National Workforce Survey.

Spring Health (2026). Survey of 1,500+ full-time employees.

Beautiful.ai (2026). AI's Impact on the Workplace: 3rd Annual Survey of American Managers.

Wellhub (2026). Return on Wellbeing Report.

Edmondson, A.C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350-383.

Eagle Hill Consulting (2025). Burnout statistics.