Middle managers and AI: closing the leadership gap

A figure that reshapes the chain of command

Two studies published in 2026 converge on a signal that organizations can hardly afford to ignore: 70% of middle managers consult generative artificial intelligence before turning to their own boss for operational advice, according to the Chartered Management Institute report signed by Salman Khalid^[1] and published on September 8, 2026. The figure arrives as Harvard Business Review^[2] describes middle managers as the pivot point that determines the actual outcome of AI adoption in organizations. The two studies tell the same story from different angles.

The 70% is a measure of actual behavior, collected in the field, and deserves to be read as such.

Two populations that remain distant

Managers who seek help from a chatbot and those who seek help from their own boss form two groups with minimal overlap. Those who turn to AI seek an immediate answer, available at any hour, free from hierarchical filtering. Those who turn to their boss seek context, judgment, and a political reading of the situation that no language model can truly provide.

When the second path remains difficult to pursue, the first becomes the only available option. This is the structural gap: the distance between the need for guidance and the guidance actually received.

The mechanism of the AI leadership tax

Khalid describes this phenomenon as a genuine tax: the time, judgment, and emotional energy that managers invest in making AI work, added to everything they were already doing. The technology promises efficiency, yet leaves many managers with the feeling of working longer to achieve the same result. The mechanism emerges when a new tool enters a process that no one has redesigned around it.

Middle managers remain the point where senior strategy meets the daily work of people. They translate vision into action, answer difficult questions, support teams through uncertainty, and ensure that daily performance holds steady while new ways of working are introduced.

Why AI amplifies uncertainty instead of reducing it

A typical middle manager must understand emerging tools, identify where they add value, question automatically generated output, and ensure responsible technology use. All of this adds to people management, performance care, team wellbeing, and customer demands.

When leadership above them remains vague about direction, AI becomes the fastest channel to fill that void. The result is paradoxical: a tool designed to lighten the load ends up multiplying it, precisely for those who must translate strategy into concrete action every day.

What high-performing leaders do differently

The evidence collected from both studies points to a recurring pattern among organizations that narrow this gap.

• They build regular discussion moments between middle managers and senior leadership, explicitly dedicated to AI-related decisions
• They clarify which judgments remain the domain of humans and which can be delegated to tools
• They invest in targeted training for middle managers, recognizing them as critical translators of transformation
• They measure the real workload perceived by managers, rather than limiting themselves to measuring tool adoption

These behaviors convert a leadership void into a structured relationship. The talent already present in the organization, with its knowledge of processes and internal relationships, responds better to this kind of intervention than to solutions imported from outside.

What this evidence means for those leading the organization

For the CEO, the 70% figure is material to bring to the board as an indicator of organizational readiness, distinct from tool adoption numbers. For the CHRO, it points to a clear development priority: L&D programs built around the middle management role, not just technical literacy. For the CFO, the leadership tax described by Khalid represents a hidden cost that justifies targeted investments in training, with measurable return on team productivity.

For the board's talent and compensation committee, the metric to monitor becomes the frequency with which middle managers find answers from their own leadership, before the frequency with which they use AI. These are two distinct signals, and they should be kept separate in reading internal reports.

The design question that remains open

Digital transformation continues to be presented as a technology project. The evidence collected in 2026 tells a different story: the 70% of managers who prefer a chatbot to their own boss is a figure about who leads, before it is about who works.

The question that every organization should ask itself concerns the quality of guidance available at middle levels, now, at this moment of transformation. Practical answers already exist, and have been observed, documented, and measured: what remains is to decide who has the will to apply them.