Leaders often ask managers to become the local translators of AI adoption. Managers are expected to encourage experimentation, explain expectations, model appropriate use, recognize good judgment, and intervene when AI-assisted work creates risk. Those responsibilities make sense. The problem is that leaders may evaluate whether managers are providing support by asking the managers themselves, while employees are making a different assessment based on what they can actually see and use in the work.

That distinction matters because support is not an intention. It is an operating condition. A manager may believe they are encouraging AI use because they mentioned it in a team meeting, forwarded a policy, or invited employees to experiment. An employee facing a consequential task still needs to know which uses are appropriate, how quality will be judged, when additional review is required, and whether questioning an AI-assisted output will be treated as good judgment or reluctance.

The Support Gap Is Not Just a Communication Problem

Charter's October 2026 AI Pulse, based on a survey of 5,002 US desk workers, managers, and executives who use AI plus 91 interviews, found a substantial difference between what managers reported doing and what employees recalled seeing. Charter described a 23-to-30 percentage-point demonstration gap across recommended manager practices. It also reported that 11 of the 14 factors most associated with employees benefiting from AI involved manager actions.

The study is cross-sectional and based largely on self-report, so it does not prove that a particular management practice caused a better outcome. Managers and employees may also interpret the same action differently. Even with those limitations, the gap raises an important organizational question: if support is visible to the manager but not to the employee, is it functioning as support?

The answer cannot be resolved by sending another message that managers are expected to help their teams. The organization needs to understand what employees require at the point of work and whether managers have the authority, capability, and time to provide it. Otherwise, the organization may have a leadership expectation without an operating practice.

AI Support Becomes Real in the Work

Visible support is not a manager repeatedly announcing enthusiasm for AI. It is the set of recognizable practices that reduces ambiguity while the work is being performed. Employees should be able to identify what good AI-assisted work looks like in their role, where human judgment must remain explicit, which outputs require review, and how to raise a concern without being penalized for slowing the process.

Those practices need to be concrete enough to guide decisions. A manager might explain why AI is appropriate for an early draft but not for a final recommendation, identify the evidence that must be checked before an output is used, or clarify who owns a decision when an AI-generated analysis conflicts with professional judgment. The organization may also need to distinguish experimentation from production use, particularly when customer information, employment decisions, financial consequences, or regulated work are involved.

When these expectations remain implicit, employees have to infer them. Some will use AI cautiously and appear slow. Others will optimize for visible speed or output. Still others may conceal their use because they are unsure how it will be judged. The resulting variation can look like uneven capability or resistance even when employees are responding rationally to unclear conditions.

Manager Intent Cannot Compensate for Missing Systems

Other research points in the same general direction while using different samples and measures. Gallup's 2026 workplace AI indicator reports that 36% of employees in organizations integrating AI strongly agreed that their manager actively supported the team's AI use. Employees who strongly agreed were more likely to report frequent use and positive effects. Microsoft's 2026 Work Trend Index, based on 20,000 surveyed AI users across 10 countries, reported that organizational factors such as culture, manager support, and talent practices accounted for more of the variation in self-reported AI impact than individual factors in its analysis.

These findings should not be treated as proof that visible manager behavior alone will produce adoption or value. Supportive managers may work in organizations with better tools, clearer strategy, stronger learning resources, or more mature workflows. Employees may also need decisions that a line manager cannot make, such as changes to performance measures, staffing, data access, job design, or formal review rights.

That is why assigning responsibility to managers without changing the surrounding system can create another burden rather than a solution. Managers may be asked to encourage AI use while lacking clarity about quality standards, risk ownership, workload expectations, or what happens to time released through automation. Their own uncertainty can remain invisible until employees encounter inconsistent guidance or a consequential error.

Visibility Is Not Performative Leadership

Making support visible does not require leaders to turn every manager into a public AI advocate. It requires alignment between what the organization expects, what managers can authorize, and what employees experience. A manager who uses AI constantly is not necessarily providing better support than one who establishes careful boundaries and helps the team make sound decisions.

The relevant evidence is not how often managers mention AI. It is whether employees can point to practices that help them do the work. Those practices may include agreed quality criteria, examples of appropriate and inappropriate use, protected time for hands-on learning, a reliable review path, and explicit permission to challenge an output. They should reflect the actual responsibilities and risks of the role rather than an organization-wide usage target.

This also means that leaders should not interpret low employee recall as proof that managers failed to communicate. The underlying problem may be a lack of decisions for managers to communicate. If the organization has not resolved how AI-assisted work will be evaluated, where accountability sits, or what employees can contest, managers cannot make those conditions real through enthusiasm alone.

What Leaders Need to Examine

Before concluding that managers are supporting AI adoption effectively, executives should compare management intent with employee experience and the observable work environment. Useful questions include:

  • Can employees describe what appropriate AI use looks like in their specific work?
  • Do managers and employees agree on the quality, review, and disclosure expectations that apply?
  • Can managers authorize experimentation, change a workflow, or resolve a conflict when existing measures discourage the new behavior?
  • Is there a clear route for questioning an AI-assisted output or escalating a consequential decision?
  • Are managers receiving the time, guidance, and authority required to provide the support leadership expects?
  • Does the organization recognize careful restraint and verification as evidence of capability when they produce a better outcome?

These questions move the discussion from whether managers support AI in principle to whether support is available when employees need to make a decision. They also help distinguish a management-development need from a broader operating-model problem.

The Larger Organizational Consequence

Organizations can report that leaders are aligned, managers have been briefed, and employees have access to approved tools while the daily work remains uncertain. That creates a false sense of readiness. The organization sees the support it intended to provide, while employees experience the decisions that were never made.

AI adoption becomes more credible when support is observable in the work: expectations are role-relevant, review is usable, judgment is protected, and managers can act on the issues employees raise. If employees cannot see or use the support, leaders should not assume the problem is visibility alone. The more consequential question is whether the organization has created anything concrete for managers to demonstrate.

References

1. Charter, "AI Pulse", October 21, 2026. 2. Gallup, "Artificial Intelligence" workplace indicator, updated April 2026 with May 2026 measures. 3. Microsoft, "2026 Work Trend Index Annual Report", May 5, 2026.

AI-assistance disclosure: This article was developed with AI assistance and substantively reviewed, revised, and approved by Change Collaboration. The final content, source use, professional judgment, and conclusions remain Change Collaboration's responsibility.