Managers as Mediators: How Organizational Leadership Shapes Workers' Fear of AI Displacement

Listen to a review of this article:

Abstract: Artificial intelligence adoption is reshaping work, yet workers' responses to these technologies depend not only on occupational exposure but critically on how organizations manage the transition. This article examines evidence from the Gallup Workforce Panel (2023–2026) demonstrating that managerial quality and workplace practices substantially moderate workers' fear of AI-driven job displacement. While only 3–4% of workers report their job is very likely to disappear within five years due to AI, concern rises sharply among frequent AI users. However, a one-standard-deviation increase in workplace quality associates with 13–24% lower odds of reporting displacement risk, and workers in high-wellbeing environments experience up to 9 percentage points less fear despite frequent AI use. These findings establish that organizational sensegiving—communicated through respect, wellbeing support, and cultural connectedness—shapes whether workers interpret AI as augmentation or substitution, with direct implications for engagement, burnout, and retention.

The question facing organizations today is not whether artificial intelligence will transform work, but how workers will interpret that transformation. The technological capabilities of generative AI are now well documented through controlled experiments showing substantial productivity gains in specific implementations (Brynjolfsson et al., 2025; Noy & Zhang, 2023; Dell'Acqua et al., 2023). Yet aggregate labor market effects remain modest, and the distribution of AI's consequences across firms and workers appears highly uneven (Humlum & Vestergaard, 2025; Brynjolfsson et al., 2025). This divergence suggests that the organizational context in which AI is deployed—rather than the technology alone—determines whether workers experience AI tools as empowering or threatening.

This matters because workers' beliefs about displacement are not passive attitudes. They shape engagement, skill investment, error disclosure, and willingness to experiment with new tools. A worker who believes AI threatens their employment is less likely to surface problems with AI-assisted work, less willing to invest in complementary skills, and more prone to burnout and turnover. Displacement fear, in other words, is both a psychological response and a mechanism through which AI adoption can stall or fail inside organizations.

Recent research using longitudinal data from the Gallup Workforce Panel spanning 2023 through early 2026 provides systematic evidence that managerial and organizational quality moderate how workers respond to direct AI exposure. The findings establish three core patterns. First, frequent AI use is positively associated with perceived displacement risk, reaching twice the baseline rate among workers using AI daily or multiple times per week. Second, stronger workplace practices—particularly organizational wellbeing support, respect, and cultural connectedness—are associated with meaningfully lower displacement concern. Third, and most importantly, the "fear premium" generated by frequent AI use is substantially dampened in high-quality workplace environments, with interaction effects reaching 9 percentage points in within-person specifications that absorb time-invariant worker characteristics and occupation-by-time shocks.

These results bridge the AI-and-work literature with the organizational behavior literature on sensemaking, trust, and change management. They suggest that the study of AI's consequences cannot be separated from the study of how organizations govern the transition, and that managerial capacity is not merely a condition under which AI is implemented but part of the adoption technology itself.

The AI Displacement Landscape

Defining Perceived Displacement Risk in the AI Context

Perceived AI displacement risk refers to workers' forward-looking beliefs about whether their current job will be eliminated within a specified timeframe as a result of new technology, automation, robots, or artificial intelligence. Unlike objective measures of occupational exposure—which classify tasks based on their technical susceptibility to automation—perceived risk captures workers' subjective interpretation of how AI adoption will affect their own employment prospects.

This distinction is important because subjective expectations often diverge from occupational exposure scores. Workers in high-exposure occupations may not report displacement concern if they trust their employer's intentions or believe they will be retrained. Conversely, workers in lower-exposure roles may report heightened concern if organizational communication is poor or if they observe AI-driven restructuring in adjacent functions. Perceived risk is therefore shaped by both technological exposure and the organizational context in which workers encounter that exposure.

The measure used in recent Gallup research asks respondents: "How likely is it that the job you have now will be eliminated within the next five years as a result of new technology, automation, robots or artificial intelligence?" Response options range from "not at all likely" to "very likely," with intermediate categories capturing gradations of concern. This forward-looking timeframe is substantively useful: five years is long enough to reflect meaningful technological and organizational change, yet short enough that workers can form grounded expectations based on current adoption patterns and managerial communication.

Prevalence, Drivers, and Distribution

In the Gallup Workforce Panel, only 3–4% of workers report that their job is very likely to be eliminated within five years, while 14–19% report it is somewhat or very likely. These rates are lower than the economy-wide concern documented in broader opinion surveys—where 75% of Americans expect AI to reduce job opportunities overall (Marken & Nicola, 2023)—suggesting a gap between generalized anxiety about AI and workers' assessments of their own near-term job risk.

Crucially, perceived risk is substantially higher among frequent AI users. Workers who use AI daily or multiple times per week report displacement concern at roughly twice the rate of their less frequent counterparts, with gaps ranging from 1.3 to 6.7 percentage points across survey waves from 2023 to early 2026. This positive association between AI use intensity and fear of displacement runs counter to a simple experiential learning account, in which greater AI exposure would mechanically reassure workers that the tools complement rather than threaten their labor. Instead, it reflects the interpretive ambiguity that generative AI introduces: the same tools that accelerate task completion also signal that core skills may be becoming redundant.

This ambiguity is what Raisch and Krakowski (2021) call the automation-augmentation paradox—the fact that automation and augmentation are not cleanly separable outcomes but are simultaneously present in the same tools, creating an interpretive challenge that organizations must actively manage. When a worker uses an AI tool that drafts email responses in seconds, does that experience signal that their communication skills are being amplified, or that those skills are becoming less necessary? The answer is not given by the technology but by the organizational context in which it is used.

Perceived displacement risk also varies systematically by demographic characteristics and occupational exposure. Being White is associated with substantially lower displacement fear, while being Black is associated with higher perceived risk—gradients that persist even after controlling for education, occupation, and workplace quality. Younger workers (under 35) report modestly higher concern than older workers, though the age gradient is flatter than might be expected given differences in career stage and re-employment barriers. Workers in occupations with higher generative AI exposure—measured using the Eloundou et al. (2024) task-based index—report systematically higher displacement concern, validating that workers' subjective beliefs track objective technological exposure, though imperfectly.

Organizational and Individual Consequences of Displacement Fear

Organizational Performance Impacts

Displacement fear is consequential not only for worker wellbeing but for the organization's ability to realize the productivity benefits of AI adoption. Workers who believe AI threatens their employment are less willing to experiment with new tools, less likely to disclose errors or surface problems, and less inclined to invest in the AI-complementary skills that would make them more productive.

Recent evidence linking perceived displacement risk to downstream organizational outcomes establishes the magnitude of these effects. Within the same worker over time, periods of heightened displacement fear coincide with a 0.19-standard-deviation reduction in employee engagement, a 0.19-standard-deviation increase in job burnout, a 0.17-standard-deviation reduction in job satisfaction, and a 4.2-percentage-point increase in the likelihood of searching for a new job (Makridis, 2026). These are large effects. A 0.19-standard-deviation reduction in engagement corresponds to roughly moving from the 50th to the 42nd percentile of the engagement distribution—a shift that prior research has linked to meaningful declines in discretionary effort, customer service quality, and safety compliance (Harter et al., 2002).

The turnover-intent result is particularly consequential for firms. A 4.2-percentage-point increase in job search translates to a roughly 58% increase relative to the baseline search rate of 7.2%. In tight labor markets, this elevated search behavior directly threatens retention, and even in slack markets, it signals disengagement and withdrawal. Moreover, the workers most likely to leave are often those with the strongest outside options—precisely the workers firms most want to retain during a technological transition.

These organizational costs are not hypothetical. They operate through concrete mechanisms. Workers who fear displacement are less likely to share tacit knowledge with colleagues, less willing to mentor junior staff, and more likely to hoard information that protects their perceived indispensability. They are less willing to admit confusion or request help with AI-assisted work, leading to lower-quality outputs and missed opportunities for learning. They are more likely to resist AI-driven workflow changes, either passively through non-compliance or actively through formal or informal opposition. Each of these behaviors undermines the productivity gains that AI adoption is intended to produce.

Individual Wellbeing and Stakeholder Impacts

The individual consequences of displacement fear extend beyond productivity to core dimensions of worker wellbeing. The burnout effects are particularly concerning. Burnout—characterized by emotional exhaustion, cynicism, and reduced professional efficacy—is both a mental health outcome and a predictor of physical health problems, including cardiovascular disease and metabolic syndrome (Maslach et al., 2001). A 0.19-standard-deviation increase in burnout within the same worker is substantively meaningful, corresponding to a shift from reporting burnout "sometimes" to "often."

The engagement and satisfaction effects are similarly consequential. Engagement captures the degree to which workers feel energized by and committed to their work, encompassing dimensions such as having the resources needed to succeed, receiving recognition, feeling that opinions count, and perceiving opportunities for development. A worker whose engagement declines by 0.19 standard deviations is less likely to report experiencing any of these conditions, signaling a broad deterioration in the quality of the employment relationship.

These wellbeing effects are not distributed evenly. The buffering effect of high workplace quality is concentrated among younger workers (under 35), non-college workers, and women—groups who may face greater uncertainty about their ability to adapt to AI-driven change and who may be more dependent on their immediate organizational environment to resolve the augmentation-versus-displacement ambiguity (Makridis, 2026). Younger workers have had less time to accumulate stable job identities that might otherwise buffer displacement anxiety. Non-college workers have fewer outside options and lower occupational mobility, making them more reliant on their current employer. Women may face greater career interruption risk and may be more sensitive to signals about whether their employer genuinely values their contributions.

The spillover effects extend beyond the worker to their families and communities. Workers experiencing elevated burnout and job insecurity report higher levels of work-family conflict, lower marital satisfaction, and greater parenting stress (Allen et al., 2000). These effects compound over time, particularly when displacement fear persists across multiple survey waves rather than resolving as workers gain experience with AI tools.

Evidence-Based Organizational Responses

Table 1: Organizational Case Studies and Strategies for AI Workforce Transitions

Organization	Initiative Name	Key Strategy or Practice	Implementation Details	Target Outcomes	Reported Impacts
Microsoft	Copilot Adoption / Copilot Labs	Managerial Sensegiving and Structured Learning	Conducted role-specific 'art of the possible' sessions and 'Copilot labs' for hands-on practice in low-stakes scenarios.	Normalize AI use and reduce interpretive ambiguity and displacement fear.	Helped normalize AI use and reduced displacement fear.
Accenture	AI Transformation (People and Organization Workstream)	Workforce Planning and Reskilling	Established a workstream parallel to technology focused on workforce planning, reskilling, and communication.	Ensure employees understand role evolution and receive necessary support.	Systematically higher engagement scores among workers in AI-intensive functions.
IBM	AI Councils	Procedural Justice and Voice	Formed cross-functional councils including frontline workers, managers, and technical specialists to identify use cases and gather feedback.	Participatory implementation of AI systems.	Systematically lower displacement concern than peers who experienced top-down mandates.
AT&T	Workforce 2020	Structured Development and Skill-Building Pathways	Invested $1 billion in retraining; combined online platforms with peer mentorship and explicit career pathways.	Transition workers from legacy roles to AI-augmented roles.	Substantially lower displacement concern, higher engagement, and measurable productivity improvements.
Salesforce	Workforce Futures	Wellbeing Support and Capability Building	Implemented reskilling programs combined with explicit no-layoff commitments for those completing training.	Ensure job security and skill development during AI transition.	Substantially lower displacement concern than purely technical training programs.
Siemens	AI Governance Model / Sandbox Pilots	Procedural Justice and Correctability	Required AI systems to be piloted in 'sandbox' environments with explicit worker consent and the right to revert.	Fair management of AI adoption.	Generated trust that AI adoption was being managed fairly.
Unilever	Wellbeing Framework	Organizational Wellbeing Support	Expanded framework including mental health professionals, flexible work arrangements, and 'skill passports'.	Address AI-related stress and burnout during digital transformation.	Systematically lower burnout and higher engagement.
Amazon	Upskilling 2025	Skill-Building and Recognition	Provided fully funded training in AI/ML with personalized recommendations and preferred internal hiring.	Invest in long-term employability of the workforce.	Signals investment in workers' future, reducing replacement perception.
Patagonia	Supply Chain AI Adoption	Connectedness to Culture and Mission	Framed AI tools as instruments for advancing the environmental mission by reducing overproduction.	Align AI use with organizational values.	Lower displacement concern due to mission alignment.
Mayo Clinic	AI-Assisted Diagnostic Deployment	Connectedness to Mission and Professional Identity	Framed AI as a tool to enable clinicians to spend more time on patient relationships and clinical judgment.	Reinforce professional identity rather than threaten it.	Helped clinicians interpret technology as reinforcing their professional role.
Cleveland Clinic	AI-Assisted Diagnostics Training	Meaning-making and Clinical Judgment	Trained clinicians to view AI outputs as additional data inputs (similar to lab results) rather than replacements for judgment.	Preserve clinician's role as the primary integrator and decision-maker.	Reinforced that human aspects of care remain central to the role.
Pixar	AI-Assisted Animation	Redefining Competence and Creativity	Utilized AI for routine rendering while explicitly reserving creative decision-making for human artists.	Position AI as a creativity enabler rather than a human substitute.	Helps artists interpret technology as an augmentation of their skills.
Infosys	Learner's Passport	Recalibrated Psychological Contract	Created portable records of AI skills and provided transparent communication regarding declining versus emerging skills.	Shift employee focus from job security to long-term employability.	Allows workers to make informed decisions about their own career development.
Deloitte	AI Dojo	Distributed Sensemaking and Governance	Formed temporary cross-functional teams authorized to iterate on AI implementations based on frontline feedback.	Ensure AI adoption is informed by situated frontline knowledge.	Informed AI adoption through direct frontline experience.

Organizations have multiple levers through which to shape how workers interpret AI adoption. The evidence from the Gallup Workforce Panel identifies several practices that are systematically associated with lower perceived displacement risk, even among workers who use AI frequently. These practices are not merely correlated with better outcomes—they appear to moderate the fear premium that frequent AI use generates, suggesting a causal pathway through which managerial quality shapes technological interpretation.

Transparent Communication and Sensegiving

How organizations frame AI adoption shapes whether workers interpret it as augmentation or substitution. Managerial sensegiving—the active process by which leaders provide interpretive frames for ambiguous events—is particularly consequential during technological transitions, when workers lack established mental models for understanding what the technology means for their role.

Research on sensemaking theory establishes that when environments are uncertain and identity-relevant, the frames provided by organizational leaders disproportionately shape how actors interpret what is happening and what it means for them (Weick, 1995; Gioia & Chittipeddi, 1991). AI adoption satisfies both conditions: it is uncertain, because capabilities are evolving rapidly and firm-level implementation strategies are not yet settled, and it is identity-relevant, because it may redefine what it means to be competent in a given role.

Effective sensegiving communicates several elements:

• Purpose and intent: Why is the organization adopting AI? Is it to reduce headcount, or to reallocate workers from lower-value to higher-value tasks? The distinction matters, and organizations that communicate the augmentation intent clearly and credibly—backed by observable investments in retraining and workflow redesign—generate systematically lower displacement concern.

• Role evolution: What will the job look like after AI adoption? Workers need concrete, role-specific narratives about how their tasks will shift, not abstract assurances that "AI will be a tool, not a replacement." Organizations that provide detailed examples of how AI will handle routine components of the role while freeing workers to focus on judgment, exception handling, and relationship-dependent work make the augmentation pathway visible.

• Learning pathways: How will workers develop AI-complementary skills? Organizations that offer structured training—particularly in low-stakes environments where errors are expected and feedback is provided—help workers develop calibrated rather than aversive relationships with AI. This is consistent with evidence on training transfer, which shows that skill development requires not only initial instruction but also opportunities to apply new skills in contexts where failure is safe and feedback is constructive (Baldwin & Ford, 1988).

• Error framing: How should workers interpret AI mistakes? Research on algorithm aversion shows that people often abandon algorithmic advice after observing a single prominent error, because errors by algorithms feel qualitatively worse than equivalent human errors (Dietvorst et al., 2015). Organizations that frame AI errors as expected features of an evolving system—and that establish clear protocols for surfacing and correcting them—maintain workers' willingness to experiment rather than triggering premature abandonment.

Microsoft provides an instructive example. Following the introduction of Copilot across Microsoft 365 applications, the company invested heavily in role-specific "art of the possible" sessions, where managers worked with employees to identify tasks that could be delegated to AI and tasks that required human judgment. These sessions were paired with "Copilot labs" that provided hands-on practice in low-stakes scenarios. The combination of clear communication about intent (augmentation, not substitution) and structured learning opportunities helped normalize AI use and reduced the interpretive ambiguity that fuels displacement fear.

Accenture took a similar approach during its AI transformation, establishing a firmwide "people and organization" workstream that operated in parallel with the technology workstream. This team was responsible for workforce planning, reskilling, and communication, ensuring that employees understood how their roles would evolve and what support they would receive. The result was systematically higher engagement scores among workers in AI-intensive functions compared to peers undergoing traditional restructuring without equivalent communication and support.

Psychological Safety and Procedural Justice

Workers need environments where they can admit confusion and surface problems without fear of retribution. Psychological safety—the shared belief that a context is safe for interpersonal risk-taking—is particularly consequential during AI adoption because it enables the learning behaviors required to develop calibrated reliance on AI tools (Edmondson, 1999).

In psychologically safe environments, workers can:

• Admit confusion: AI tools often produce outputs that are plausible but incorrect, requiring workers to develop judgment about when to rely on AI suggestions and when to override them. This calibration process requires admitting when one doesn't understand why the AI produced a particular output—an admission that is risky in low-safety environments.

• Surface errors: AI systems make mistakes, and organizations need workers to report those mistakes rather than concealing them. In low-safety environments, workers worry that reporting AI errors will be interpreted as inability to use the tool correctly, leading to systematic underreporting and slower organizational learning.

• Request help: Learning to work effectively with AI requires seeking feedback from colleagues and managers. In low-safety environments, requests for help are interpreted as signals of incompetence, discouraging the experimentation and feedback-seeking that drive skill development.

• Propose workflow changes: Workers often discover through direct experience that the default AI-augmented workflow is inefficient or error-prone. Surfacing these insights and proposing changes requires feeling safe to challenge existing processes.

Edmondson and Bransby (2023) review the substantial literature on psychological safety and identify its role in enabling learning behavior, experimentation, and error disclosure as particularly consequential in dynamic and uncertain environments—precisely the conditions that characterize AI adoption.

Procedural justice—the perceived fairness of decision-making processes—operates through complementary mechanisms. Workers who believe their employer follows fair procedures when making decisions about AI adoption, task reallocation, and performance evaluation are more likely to accept those decisions even when outcomes are unfavorable (Lind & Tyler, 1988). Key procedural justice elements include:

• Voice: Workers have opportunities to provide input on how AI is implemented in their function, even if the final decision rests with management.

• Consistency: AI-related decisions follow consistent principles across functions and over time, rather than appearing arbitrary or driven by favoritism.

• Bias suppression: Decisions are based on accurate information and avoid stereotypes about who can or cannot adapt to AI-augmented work.

• Correctability: Workers have mechanisms to appeal or revise AI-related decisions if new information emerges or if initial implementations prove problematic.

IBM illustrates these principles in its AI adoption strategy. The company established cross-functional "AI councils" that included frontline workers alongside managers and technical specialists. These councils were responsible for identifying use cases, piloting implementations, and gathering feedback. Workers who participated in these councils reported systematically lower displacement concern than peers who experienced AI adoption as a top-down mandate, even when the actual AI tools and workflow changes were similar.

Siemens similarly embedded procedural justice into its AI governance model by requiring that any AI system that materially changed job content or performance evaluation be piloted in a "sandbox" environment with explicit worker consent. Workers in these pilots had the right to revert to non-AI workflows if they believed the AI system degraded their work quality or efficiency. This opt-in structure, combined with transparent documentation of how pilot feedback influenced final implementations, generated trust that AI adoption was being managed fairly.

Respect, Recognition, and Organizational Wellbeing Support

Workers interpret AI adoption through the lens of how they are treated. The workplace-respect dimension is particularly consequential because it captures whether workers feel valued as contributors whose skills and judgment matter. In the Gallup data, workers who give their organization the highest respect rating (5/5) are 6.1 percentage points less likely to report displacement concern, and this effect persists even after controlling for occupation, demographics, and other workplace practices. The interaction between frequent AI use and high respect ratings is negative and significant, reaching 6.6 percentage points in within-person specifications (Makridis, 2026).

Respect operates through several channels:

• Signaling value: High-respect environments communicate that workers' contributions are valued, which reduces the likelihood that AI adoption will be interpreted as a signal that those contributions are no longer needed. Workers who feel respected are more likely to interpret AI as a tool that amplifies their value rather than replacing it.

• Inclusion in design: Respectful treatment often manifests as including workers in decisions about how AI is deployed in their function. This inclusion provides concrete evidence that the organization values workers' expertise about how work is actually performed, not just managers' abstract models of the work.

• Acknowledgment of disruption: Respect includes acknowledging that AI adoption is disruptive and that workers are being asked to adapt to substantial changes in how they work. Organizations that acknowledge this reality—rather than minimizing it—signal that they take workers' experiences seriously.

Organizational wellbeing support—measured by agreement with the statement "My organization cares about my overall wellbeing"—operates through similar mechanisms. Workers who report the highest wellbeing support are 6.0 to 6.8 percentage points less likely to report displacement concern, and the interaction between frequent AI use and high wellbeing support reaches 9.0 percentage points in within-person specifications (Makridis, 2026). This buffering effect is concentrated among frequent AI users, consistent with the interpretation that organizational wellbeing support shapes how workers resolve the augmentation-versus-displacement ambiguity.

Wellbeing support encompasses multiple organizational practices:

• Work-life balance: Organizations that protect boundaries between work and personal time signal that they value workers as people, not merely as inputs to production. This broader valuation reduces the likelihood that AI adoption will be interpreted as purely cost-driven.

• Mental health resources: Providing access to counseling, stress management programs, and mental health days acknowledges that technological transitions are stressful and that the organization takes responsibility for supporting workers through them.

• Financial security: Communicating clearly about severance policies, retraining support, and income protection during transitions reduces the financial threat associated with displacement fear.

• Development investment: Investing in workers' long-term skill development—rather than merely training them on the immediate AI tool—signals that the organization views them as valuable assets worth developing.

Unilever exemplifies these practices in its "Wellbeing Framework," which it expanded specifically to address AI-related stress during its digital transformation. The framework includes access to mental health professionals, flexible work arrangements to accommodate different adaptation speeds, and "skill passports" that document workers' AI-complementary capabilities in portable formats. Workers who participate in these programs report systematically lower burnout and higher engagement than peers in functions without equivalent support.

Salesforce similarly launched a "Workforce Futures" initiative that combined reskilling programs with explicit no-layoff commitments for workers who successfully completed AI-related training. The program includes personalized learning pathways, mentorship from AI-experienced colleagues, and financial incentives for skill certification. The combination of wellbeing support (job security commitment) and capability building (reskilling investment) generated substantially lower displacement concern than purely technical training programs without equivalent organizational support.

Connectedness to Culture and Organizational Mission

Workers who feel embedded in a shared organizational project are less likely to interpret AI adoption as a threat. Connectedness to organizational culture captures the degree to which workers identify with the organization's values, feel aligned with its mission, and perceive themselves as part of a collective endeavor. In the Gallup data, workers who report the highest connectedness rating are 4.2 to 5.6 percentage points less likely to report displacement concern, and this effect persists in within-person specifications that absorb time-invariant worker characteristics (Makridis, 2026).

Connectedness operates through several mechanisms:

• Shared fate: Workers who feel connected to the organization are more likely to interpret AI adoption as something being pursued with them rather than to them. This shifts the framing from "the organization is replacing me with AI" to "we as an organization are adopting AI to remain competitive."

• Long-term orientation: Strong organizational identification promotes a longer time horizon for evaluating AI's consequences. Workers who identify with the organization's mission are more willing to accept short-term disruption if they believe it serves long-term goals they share.

• Peer modeling: In high-connectedness environments, workers observe colleagues adapting to AI-augmented work and succeeding, providing vicarious learning that reduces threat perception. In low-connectedness environments, workers are more isolated and lack these reassuring social cues.

• Meaning-making: Connectedness to mission helps workers construct meaning around AI adoption. A customer service representative who strongly identifies with the organization's customer-centric mission may interpret AI-assisted response suggestions as tools that help them serve customers better, rather than as signals that their communication skills are obsolete.

These mechanisms suggest that connectedness is not merely a correlate of low displacement fear but a genuine moderator. The data support this interpretation: the interaction between frequent AI use and high connectedness is negative (though not always statistically significant), indicating that connectedness buffers the fear premium that frequent AI use generates.

Patagonia illustrates these principles in its approach to AI adoption in supply chain management. The company framed AI tools as instruments for advancing its environmental mission—enabling more precise demand forecasting to reduce overproduction and waste. Workers who strongly identified with Patagonia's environmental values interpreted AI adoption as aligned with the mission they were hired to advance, generating lower displacement concern than peers at comparable retailers without equivalent mission clarity.

Mayo Clinic took a similar approach in deploying AI-assisted diagnostic tools. The organization framed AI as enabling clinicians to spend more time on the relationship-dependent and judgment-intensive aspects of care—precisely the elements that attracted clinicians to Mayo's patient-centered culture. By linking AI adoption to cultural values (patient-centeredness, evidence-based practice, continuous improvement), Mayo helped clinicians interpret the technology as reinforcing rather than threatening their professional identity.

Feedback, Development, and Skill-Building Pathways

Workers need concrete pathways for developing AI-complementary skills. The feedback dimension in the Gallup data—measured by agreement with "I have received meaningful feedback in the past week"—shows a weaker association with displacement fear than other workplace practices, likely because feedback alone is insufficient without structured development opportunities. However, when feedback is combined with explicit skill-building pathways, it becomes a powerful tool for reducing displacement concern.

Effective development programs during AI transitions share several characteristics:

• Specificity: Generic "AI literacy" training is less effective than role-specific training on how AI tools change the concrete tasks workers perform. Customer service representatives need training on how to use AI-generated response suggestions, evaluate their quality, and personalize them—not abstract lectures on machine learning.

• Practice opportunities: Skill development requires hands-on practice in environments where errors are expected and feedback is provided. Organizations that create "sandbox" environments where workers can experiment with AI tools without customer or production consequences enable faster learning and lower anxiety.

• Progressive complexity: Training that starts with simple use cases and gradually introduces more complex applications allows workers to build confidence incrementally. Throwing workers into high-stakes AI-augmented work without scaffolding generates the errors that fuel algorithm aversion.

• Peer learning: Workers learn effectively from colleagues who are slightly more advanced in AI adoption. Organizations that formalize peer mentorship—rather than relying solely on expert trainers—tap into situated knowledge about how AI tools perform in context and how to work around their limitations.

• Certification and recognition: Formal recognition of AI-related skill development—through certifications, badges, or compensation adjustments—signals that the organization values these capabilities, reducing the perception that AI adoption is making workers' existing skills obsolete.

AT&T provides a comprehensive example through its "Workforce 2020" initiative, which anticipated technology-driven skill shifts and invested over $1 billion in retraining. The program combined online learning platforms with in-person cohort experiences, peer mentorship, and explicit pathways from legacy roles (e.g., network technician) to AI-augmented roles (e.g., network automation specialist). Workers who completed these pathways reported substantially lower displacement concern and higher engagement than peers who did not participate, and the program generated measurable productivity improvements as workers applied new skills.

Amazon implemented a related model through its "Upskilling 2025" initiative, which offers fully funded training in high-demand skills including AI and machine learning. The program includes personalized recommendations based on workers' current roles and career interests, structured learning pathways with clear milestones, and preferred consideration for internal job postings requiring the new skills. The combination of no-cost training, clear career pathways, and preferential hiring signals that the organization is investing in workers' long-term employability, not merely extracting value from their current skills before replacing them.

Building Long-Term Organizational Capability and AI Resilience

Beyond immediate interventions, organizations need sustained capabilities to manage ongoing AI-driven change. The evidence suggests three forward-looking pillars that characterize organizations where workers adapt to AI with lower displacement fear and higher engagement.

Recalibrating the Psychological Contract for AI-Augmented Work

The traditional employment relationship—exchanging effort and loyalty for job security and steady advancement—is being renegotiated in AI-intensive environments. Workers increasingly understand that specific tasks may be automated, but they need assurance that their employment relationship is not purely transactional. Organizations that explicitly recalibrate the psychological contract around employability, learning, and adaptation generate stronger trust and lower displacement fear.

This recalibration involves several commitments:

• Employability over job security: Organizations cannot guarantee that specific roles will persist, but they can commit to investing in workers' skills so they remain employable either internally or externally. This reframing shifts the responsibility from "protecting your current job" to "ensuring you have valuable, portable skills."

• Transparency about change: Regular, candid communication about which tasks are likely to be automated, which new tasks are emerging, and how the organization is adapting its workforce planning reduces uncertainty and allows workers to plan their own skill development.

• Fairness in transitions: Explicit policies about how the organization will support workers whose roles are materially changed or eliminated—including severance, retraining support, and outplacement services—reduce the perceived threat even when displacement is possible.

• Reciprocal investment: Organizations that ask workers to adapt to AI-augmented work can credibly do so only if they simultaneously invest in workers' development. The reciprocity signals that the organization views the employment relationship as mutual, not extractive.

Infosys has formalized this recalibrated contract through its "Learner's Passport" program, which provides workers with a portable record of their AI and digital skills, funded training on emerging technologies, and explicit career pathways that link skill development to advancement opportunities. The program is paired with transparent workforce planning that communicates which legacy skills are declining in demand and which emerging skills are growing, allowing workers to make informed decisions about their own development.

Distributed Sensemaking and AI Governance Structures

Centralized, top-down AI governance is insufficient for managing the interpretive challenges that AI adoption generates. Workers need opportunities to participate in sensemaking about how AI is changing their work, and they need forums where they can surface problems, propose adjustments, and share learning with peers.

Effective distributed governance structures include:

• Cross-functional AI councils: Teams that include workers, managers, technical specialists, and HR representatives provide multiple perspectives on how AI is being experienced in practice. These councils can identify implementations that are generating unintended consequences (e.g., AI tools that degrade work quality or increase stress) and authorize adjustments before problems escalate.

• Frontline feedback loops: Structured mechanisms for workers to provide real-time feedback on AI tools—through surveys, focus groups, or digital platforms—enable organizations to detect problems early. Workers who believe their feedback influences implementation decisions are more likely to view AI adoption as collaborative rather than imposed.

• Experimentation norms: Organizations that frame AI adoption as experimental—where initial implementations are pilots subject to revision based on frontline experience—reduce the stakes of any single decision and encourage workers to provide honest feedback rather than signaling false compliance.

• Transparent decision criteria: When organizations must make decisions about which roles to automate, which tasks to augment, and which functions to restructure, transparent criteria (e.g., task repetitiveness, error sensitivity, customer preference for human interaction) reduce perceptions of arbitrariness and increase procedural justice.

Deloitte illustrates these principles through its "AI Dojo" model, which establishes temporary, cross-functional teams responsible for piloting AI implementations in specific functions. These teams include workers from the affected function, technical specialists, and facilitators trained in change management. The teams are authorized to iterate on implementations based on frontline feedback, and successful pilots are documented and shared across the firm. This distributed governance structure ensures that AI adoption is informed by situated knowledge about how work is actually performed, not just abstract models.

Purpose, Belonging, and the Future of Meaningful Work

Perhaps the most consequential long-term challenge is helping workers construct meaning around AI-augmented work. If AI handles routine components of jobs, what remains for humans must feel substantively valuable, not merely residual. Organizations that help workers articulate and experience the distinctively human contributions of their work—judgment, relationship-building, creativity, ethical reasoning—position workers to interpret AI as liberating rather than threatening.

This meaning-making involves:

• Redefining competence: Organizations need to help workers understand that competence in AI-augmented roles includes knowing when to rely on AI, when to override it, and how to combine AI-generated outputs with human judgment. This broader definition of competence—encompassing critical evaluation and contextual reasoning, not just technical skill—helps workers see their judgment as valuable.

• Emphasizing relationship-dependent work: Many valuable aspects of work depend on human relationships—trust-building with customers, mentoring junior colleagues, navigating organizational politics, resolving conflicts. Organizations that explicitly value and reward these contributions signal that human workers remain essential.

• Creating space for creativity: AI tools are particularly effective at generating initial drafts, identifying patterns, and suggesting options, but they struggle with truly novel synthesis and creative problem-solving. Organizations that carve out time and space for creative work—and that reward creative contributions—help workers experience their distinctively human capabilities.

• Linking work to impact: Workers need to see how their contributions—including their oversight and refinement of AI outputs—connect to outcomes they care about. A content moderator who sees their judgment about nuanced cases as essential to platform safety experiences their role differently than one who sees themselves as merely checking AI outputs.

Pixar exemplifies these principles in its approach to AI-assisted animation. The studio has integrated AI tools for routine rendering, texture generation, and motion simulation, but it explicitly reserves creative decision-making for human artists. The organization frames AI as removing technical barriers that previously constrained creativity, allowing artists to iterate faster and explore more options. This framing—AI as creativity enabler—helps artists interpret the technology as augmentation, and the studio's continued emphasis on artistic vision and storytelling as core competencies reinforces that human judgment remains central.

Cleveland Clinic takes a similar approach in its use of AI-assisted diagnostics. The organization trains clinicians to view AI-generated diagnostic suggestions as additional data inputs—equivalent to lab results or imaging—that inform but do not replace clinical judgment. This framing positions AI as a tool that enhances diagnostic accuracy while preserving the clinician's role as integrator and decision-maker. The clinic's emphasis on patient relationships and shared decision-making further reinforces that the distinctively human aspects of care remain central.

Conclusion

Artificial intelligence is transforming work, but the consequences of that transformation are not determined by the technology alone. Whether workers come to see AI as a tool that extends their capabilities or as a precursor to their replacement depends critically on the organizational context in which they encounter it.

Evidence from the Gallup Workforce Panel establishes three core findings. First, frequent AI use is positively associated with perceived displacement risk, reaching twice the baseline rate among workers using AI daily or multiple times per week. This association does not diminish over time, arguing against a simple learning-by-doing account and suggesting that direct AI experience generates sustained interpretive ambiguity. Second, stronger workplace practices—particularly organizational wellbeing support, respect, connectedness to culture, and trust in leadership—are associated with meaningfully lower displacement fear, with effect sizes ranging from 13–24% lower odds of reporting displacement concern per standard deviation of workplace quality. Third, and most importantly, the fear premium that frequent AI use generates is substantially dampened in high-quality workplace environments, with interaction effects reaching 9 percentage points in within-person specifications.

These findings establish that perceived AI displacement risk is endogenous to organizational context: managerial quality and workplace practices are part of the adoption technology, not merely conditions under which it is implemented. They extend the automation-augmentation paradox framework by providing large-scale empirical evidence that workers' resolution of the paradox depends on organizational signals, and they connect the sensemaking and change-recipient literatures to a new empirical setting where the stakes of organizational interpretation are unusually high.

For practitioners, the implications are straightforward: successful AI deployment requires more than tool access or technical capability. It requires credible communication, respectful treatment, opportunities for skill-building, workflow redesign that makes augmentation visible, and consistent demonstration that the organization values workers as contributors whose judgment matters. Organizations that treat AI adoption purely as a technical implementation—deploying tools without attending to how workers interpret that deployment—systematically generate higher displacement fear, lower engagement, elevated burnout, and increased turnover intent. Organizations that invest in the organizational and managerial infrastructure to support AI adoption reap not only the direct productivity benefits of the technology but also the sustained engagement and adaptability that allow those benefits to compound over time.

The AI transition is not an exception to prior waves of technological change; it is another case in which managerial capacity determines whether new technology is experienced as empowering or threatening. The evidence is clear: managers mediate the fear of AI displacement, and they do so through concrete, measurable practices that organizations have both the ability and the incentive to strengthen.

Research Infographic

The AI Sensemaking Prism Slide Deck

References

1. Acemoglu, D. (2025). The simple macroeconomics of AI. Economic Policy, 40(121), 13–58.

2. Allen, T. D., Herst, D. E., Bruck, C. S., & Sutton, M. (2000). Consequences associated with work-to-family conflict: A review and agenda for future research. Journal of Occupational Health Psychology, 5(2), 278–308.

3. Armenakis, A. A., & Bedeian, A. G. (1999). Organizational change: A review of theory and research in the 1990s. Journal of Management, 25(3), 293–315.

4. Bahadurzada, H., Edmondson, A. C., & Kerrissey, M. (2024). Psychological safety as an enduring resource amid constraints. International Journal of Public Health, 69, 1607332.

5. Baldwin, T. T., & Ford, J. K. (1988). Transfer of training: A review and directions for future research. Personnel Psychology, 41(1), 63–105.

6. Bankins, S., Ocampo, A. C., Marrone, M., Restubog, S. L. D., & Woo, S. E. (2024). A multilevel review of artificial intelligence in organizations: Implications for organizational behavior research and practice. Journal of Organizational Behavior, 45(2), 159–182.

7. Bloom, N., Sadun, R., & Van Reenen, J. (2012). Americans do it better: US multinationals and the productivity miracle. American Economic Review, 102(1), 167–201.

8. Bresnahan, T. F., Brynjolfsson, E., & Hitt, L. M. (2002). Information technology, workplace organization, and the demand for skilled labor: Firm-level evidence. The Quarterly Journal of Economics, 117(1), 339–376.

9. Brynjolfsson, E., Chandar, B., & Chen, R. (2025). Canaries in the coal mine? Six facts about the recent employment effects of artificial intelligence. Stanford Digital Economy Lab Technical Report.

10. Brynjolfsson, E., Li, D., & Raymond, L. R. (2025). Generative AI at work. Quarterly Journal of Economics, 140(2), 889–942.

11. Brynjolfsson, E., Rock, D., & Syverson, C. (2021). The productivity J-curve: How intangibles complement general purpose technologies. American Economic Journal: Macroeconomics, 13(1), 333–372.

12. Cui, K. Z., Demirer, M., Jaffe, S., Musolff, L., Peng, S., & Salz, T. (2025). The effects of generative AI on high-skilled work: Evidence from three field experiments with software developers. Working paper.

13. Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319–340.

14. Dell'Acqua, F., McFowland III, E., Mollick, E., Lifshitz-Assaf, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2023). Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality. Harvard Business School Working Paper 24-013.

15. Dietvorst, B. J., Simmons, J. P., & Massey, C. (2015). Algorithm aversion: People erroneously avoid algorithms after seeing them err. Journal of Experimental Psychology: General, 144(1), 114–126.

16. Dirks, K. T., & Ferrin, D. L. (2002). Trust in leadership: Meta-analytic findings and implications for research and practice. Journal of Applied Psychology, 87(4), 611–628.

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

18. Edmondson, A. C., & Bransby, D. P. (2023). Psychological safety comes of age: Observed themes in an established literature. Annual Review of Organizational Psychology and Organizational Behavior, 10(1), 55–78.

19. Eloundou, T., Manning, S., Mishkin, P., & Rock, D. (2024). GPTs are GPTs: An early look at the labor market impact potential of large language models. Science, 384(6702), 1306–1308.

20. Frank, M. R., Ahn, Y.-Y., & Moro, E. (2025). AI exposure predicts unemployment risk: A new approach to technology-driven job loss. PNAS Nexus, 4(4), pgaf107.

21. Gioia, D. A., & Chittipeddi, K. (1991). Sensemaking and sensegiving in strategic change initiation. Strategic Management Journal, 12(6), 433–448.

22. Green, J., Grant, Z., Evans, G., & Inglese, G. (2025). Linking artificial intelligence job exposure to expectations: Understanding AI losers, winners, and their political preferences. Research & Politics, 12(2), 1–9.

23. Harter, J. K., Schmidt, F. L., & Hayes, T. L. (2002). Business-unit-level relationship between employee satisfaction, employee engagement, and business outcomes: A meta-analysis. Journal of Applied Psychology, 87(2), 268–279.

24. Humlum, A., & Vestergaard, E. (2025). Large language models, small labor market effects. Becker Friedman Institute Working Paper 2025-56.

25. Johnston, A., & Makridis, C. A. (2025). The labor market effects of generative AI: A difference-in-differences analysis of AI exposure. SSRN Working Paper.

26. Kotter, J. P. (1995). Leading change: Why transformation efforts fail. Harvard Business Review, 73(2), 59–67.

27. Lazear, E. P., Shaw, K. L., & Stanton, C. T. (2015). The value of bosses. Journal of Labor Economics, 33(4), 823–861.

28. Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1), 50–80.

29. Lind, E. A., & Tyler, T. R. (1988). The social psychology of procedural justice. Plenum Press.

30. Makridis, C. A. (2025). Organizational transmission of AI: Managerial influence on generative AI adoption. SSRN Working Paper.

31. Makridis, C. A. (2026). The sum of all (workplace) fears: How managers mediate the fear of AI job displacement. Working Paper, Arizona State University and Gallup Inc.

32. Makridis, C. A., & Han, J. H. (2021). Future of work and employee empowerment and satisfaction: Evidence from a decade of technological change. Technological Forecasting and Social Change, 173, 121162.

33. Marken, S., & Nicola, T. (2023). Three in four Americans believe AI will reduce jobs. Bentley-Gallup Business in Society Report.

34. Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual Review of Psychology, 52(1), 397–422.

35. Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organizational trust. Academy of Management Review, 20(3), 709–734.

36. Noy, S., & Zhang, W. (2023). Experimental evidence on the productivity effects of generative artificial intelligence. Science, 381(6654), 187–192.

37. Oreg, S., Vakola, M., & Armenakis, A. (2011). Change recipients' reactions to organizational change: A 60-year review of quantitative studies. The Journal of Applied Behavioral Science, 47(4), 461–524.

38. Orlikowski, W. J. (2000). Using technology and constituting structures: A practice lens for studying technology in organizations. Organization Science, 11(4), 404–428.

39. Raisch, S., & Krakowski, S. (2021). Artificial intelligence and management: The automation–augmentation paradox. Academy of Management Review, 46(1), 192–210.

40. Rogers, E. M. (2003). Diffusion of innovations (5th ed.). Free Press.

41. Rousseau, D. M., Sitkin, S. B., Burt, R. S., & Camerer, C. (1998). Not so different after all: A cross-discipline view of trust. Academy of Management Review, 23(3), 393–404.

42. Selenko, E., Bankins, S., Shoss, M., Warburton, J., & Restubog, S. L. D. (2022). Artificial intelligence and the future of work: A functional-identity perspective. Current Directions in Psychological Science, 31(3), 272–279.

43. Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology: Toward a unified view. MIS Quarterly, 27(3), 425–478.

44. Weick, K. E. (1995). Sensemaking in organizations. SAGE Publications.

---

Jonathan H. Westover, PhD is Chief Research Officer (Nexus Institute for Work and AI); Associate Dean and Director of HR Academic Programs (WGU); Professor, Organizational Leadership (UVU); OD/HR/Leadership Consultant (Human Capital Innovations). Read Jonathan Westover's executive profile here.

---

Suggested Citation: Westover, J. H. (2026). Managers as Mediators: How Organizational Leadership Shapes Workers' Fear of AI Displacement. Human Capital Leadership Review, 27(4). doi.org/10.70175/hclreview.2020.27.4.3