AI Adoption Is Overloading Your Middle Managers

Summary:

Researchers conducted 18 semi-structured interviews with partners, managers, and junior consultants at two major consulting firms to understand how people at each level were actually using AI, what support they received, and where the friction was.

Most organizations treat AI adoption as a technology challenge—a software rollout to be managed by IT and celebrated by the C-suite. Some even see it as a fast track to headcount reduction.

To understand how AI adoption plays out in practice, we conducted 18 semi-structured interviews with partners, managers, and junior consultants at two major consulting firms. Rather than surveying broad attitudes, we asked people at each level how they were actually using AI, what support they received, and where the friction was.

What emerged was not a technology story but an organizational one. The pressure point was consistent across both firms. Our research suggests where AI adoption actually succeeds or fails: the middle layer of management.

Consider a composite portrait of a typical manager at a consulting firm:

She starts her day learning new prompting techniques before her team logs on.
She then attends her client meetings, answering questions on how to use AI or how her team is using AI for their deliverables.
By midday, she is checking AI-generated client work for errors, coaching a brand new junior analyst who has never built a deck from scratch, and trying to interpret a partner’s request for an “AI-enhanced” memo with little guidance on what that means.
At the end of the day, she documents what worked so the team can reuse its new-found learning about AI the next time.

Across interviews, versions of this story came up again and again. Our composite manager is not unusual. She may be the norm today. While our research focused on consulting, the patterns we found—managers caught between executive ambition and operational reality, with little formal support—are likely familiar to leaders across knowledge-intensive industries.

The Capability-Reality Gap

Survey data shows broad but shallow AI adoption and uneven value creation. Roughly 88% of organizations now use AI in at least one business function, but only about a quarter have developed the capabilities to generate tangible value beyond initial pilots.

McKinsey research identifies workflow redesign, not technology sophistication, as the primary driver of AI impact, and our interviews help explain why. At the senior level, leaders are leaning into AI’s strategic potential, expanding scope, accelerating delivery with leaner teams, and reimagining services. At the junior level, consultants report dramatic productivity gains: Desktop research that once took days now takes 30 minutes; analysis that once consumed weeks now only takes hours. Freed from early-stage work, juniors are doing strategic synthesis and sitting in on discovery interviews earlier in their careers than in any prior generation.

But the efficiency gains at the bottom and the strategic ambitions at the top converge on a single pressure point: middle managers. Our interviews and research reveal that managers are drowning from being responsible for catching “workslop,” AI-generated content that looks professional, but lacks substance and fails to advance the actual task. They’re expected to validate AI outputs, identify errors, coach their teams in AI skills and core on-the-job principles, and uphold quality standards—all while facing unchanged or even increased delivery pressure and lacking formal support structures.

This burden is compounding a crisis that predates AI. Middle managers were already carrying more responsibility than ever, as layoffs and reorganizations stripped away layers of support, leaving fewer people to supervise more employees. Gartner predicts that in 2026, 20% of organizations will use AI to flatten their structure, eliminating more than half of current middle management positions. And Gallup finds that manager engagement has fallen sharply from 30% in 2023 to just 22% in 2025, the steepest decline in any employee group. AI didn’t create the middle-manager burnout problem. It accelerated it.

Overloading middle managers creates a structural risk. The question facing leaders shouldn’t be whether to thin this layer but rather how to reinforce it, because when the middle layer doesn’t function well, neither the efficiency gains at the junior level nor the strategic ambitions at the partner level can translate into client value.

The AI Transition Requires Over-Investment–Especially in Middle Managers

The idea that AI can free workers for higher-value tasks is well established. What we observed was something more specific: a pattern we call role elevation. In the teams where the transition was working well, AI was being used not to eliminate roles but to shift work upward. Juniors were doing higher-value work, such as interpreting data and joining strategy conversations that would previously have been reserved for more senior staff. Partners moved from selling methodology to selling AI-enhanced judgment.

But managers did not experience role elevation. Their new responsibilities—the oversight, coaching, and quality-control demands of AI—have simply been layered onto their existing work. Without organizational support, managers don’t get elevated; they get buried.

Our interviews surfaced three distinct ways the middle layer is failing under the weight of AI adoption.

Breakdown 1: Learning is informal, while delivery is relentless.

In the firms we studied, on many teams the time AI saved was swallowed immediately by client work and delivery pressure. Managers were expected to experiment, learn, and teach others, but their formal responsibilities hadn’t changed. As a result, teams repeatedly solved the same problems. Effective prompts, workflows and governance practices remained scattered across individuals rather than being institutionalized.

Teams that handled this better consistently made two operating changes. They protected time for learning, and they made it easier for other teams to find and reuse what they had already figured out. Leadership temporarily lowered utilization targets during AI transition periods, formalized dedicated contribution time (for example, instituting weekly sessions where junior consultants shared what they learned with their teams), and tied performance reviews to how well employees document and share AI use cases. When learning time appeared on the calendar, adoption started to compound.

The differentiator between teams in our study was not which AI tools they had access to. It was whether the teams had built a centralized internal hub that consolidated tools, use cases, and governance guidance, with a robust search function so employees knew exactly where to go. The most effective AI practices often originated with frontline teams solving immediate project problems, but only scaled when the firm had infrastructure to capture and redistribute what they’d learned. The teams thus experienced less redundant experimentation and increased cross-project reuse.

Breakdown 2: Incentives reward the wrong behaviors.

AI is redefining what good performance looks like, but most evaluation systems haven’t caught up. At the firms we studied, traditional metrics still reward billable hours and individual output. Meanwhile, the behaviors that drive successful AI adoption—such as sharing effective prompts across teams, coaching others, and contributing to internal tools—go unrecognized. In some cases, we found employees avoided acknowledging AI use in their own work, reflecting an incentive structure that still equates personal effort with professional value.

The fix requires shifts at every level, but especially for managers. Senior leaders must recognize that managers are now carrying a triple burden—managing AI experimentation, maintaining client delivery, and developing people—and reward them for coaching, team development, and knowledge transfer, not just for delivery. Until incentive structures reflect that reality, managers will default to what gets measured, such as utilization percentages, and the coaching and knowledge-building that compounds adoption will be pushed to the margins.

Breakdown 3: Leaders and managers operate in different realities.

Survey data from BCG shows that executives are roughly twice as likely as individual contributors to describe employees as enthusiastic about AI. Our research confirmed this perception gap and showed where it bites hardest: at the manager level.

Partners tended to be removed from how AI changes operational work, creating a disconnect between strategic vision and day-to-day reality.The gap matters and middle managers are trying to fill it on their own. Managers decide when AI output is good enough, what juniors should still learn to do by hand, what standards to apply to client-facing work, and how to handle a client who assumes all the work they’ve received is AI-generated. When there’s no firm-wide direction or standards during this period of transition, those decisions are being made in isolation, team by team.

Closing the gap requires visible leadership engagement, framed around what managers need. When leaders joined working sessions, managers reported that the interpretive burden they’d been carrying alone began to ease; firm-wide direction replaced individual guesswork. Leaders also gained a clearer view of the practical tradeoffs managers were navigating, which helped calibrate expectations.

Simultaneously, firms need to invest directly in manager-specific AI training. Provide targeted training on AI oversight, such as hallucination detection, prompt evaluation, and fact-checking AI-generated analysis. Facilitate manager-to-manager learning forums so review techniques travel across teams rather than being invented independently. And critically, clarify firm-wide expectations for AI usage so that each manager isn’t left to interpret the rules alone.

Protecting the Pipeline

There is a deeper problem beneath the middle-manager squeeze that deserves attention: If managers are spending more time validating AI outputs and fighting fires, who is developing the next generation of leaders?

In traditional consulting, juniors learned by watching managers up close: how they structured a workplan, pressure-tested an analysis, and handled difficult client conversations. Hybrid work has already weakened some of that apprenticeship. AI may weaken it further by compressing technical tasks before judgment has been built. A junior can now produce a polished deliverable quickly. What still takes time to learn is how to tell when an analysis is plausible but weak, whether the recommendations make sense, or how to challenge a client without losing trust.

This is not just a workflow problem. It is a leadership-pipeline problem. If firms can reduce the time managers spend checking and rechecking AI output, they can redirect some of that capacity toward coaching and development. The firms that figure out how to protect that capacity will be the ones that still have a well-developed leadership pipeline in five years. The firms that don’t will discover that AI accelerated junior output but hollowed out the path from contributor to leader.

Investing in your middle layer can lead to an AI adoption that compounds across the firm. Teams build on each other’s work. Quality improves as managers develop fluency. Juniors grow faster. Partners can speak credibly about AI with clients because they’ve seen how it works inside their own teams, not just what they think could be the reality.

The difference in AI adoption isn’t about the technology. It’s whether leadership has built the support structure around the people who make AI work in practice. Three questions every leader should ask: Who in your organization is bearing the cost of your AI ambition? Are you doing enough to support them? Have you protected the capacity your managers need to develop the leaders who will follow them?

Explore AAPL Membership benefits.