Research Drop: Reframing AI Friction as a Mastery Challenge to Drive Impact

Research Drop in Brief:

• More than half of employees say agentic AI integration feels overwhelming and that they are struggling with the pace of change.
• Employees who lean into friction – navigating new AI tools even when it feels overwhelming – show up to +23pp agentic AI readiness, +18pp realized value, and are 1.9x as likely to be high-frequency AI users.
• Viewing friction as a mastery challenge versus a barrier hindrance can shape AI adoption outcomes and building trust, increasing recognition, and providing training can support this mindset shift.

Integrating AI and AI agents isn’t just a technical upgrade – it’s a cultural shift. Organizations are rethinking workflows, decision-making, and even what knowledge and expertise looks like in human-AI teams. This transition is likely to feel messy. It challenges deeply ingrained habits and mental models, creating inevitable friction. One way this friction shows up is with employees who struggle with the speed and complexity of AI adoption and integration. In our data we found that:

• 57% of employees feel overwhelmed by the idea of integrating agentic AI into their work
• 55% of employees struggle to keep up with the pace of change around agentic AI

This month’s Research Drop explores how to balance these feelings of strain with continued growth and impact.

The unexpected benefit of AI adoption friction

While we may initially assume that friction in AI transformation is something to be avoided, not all friction is equal. Technical breakdowns? Bad friction. Employees stretching beyond their comfort zone? Good friction.

Employees who are reporting this friction while navigating new AI tools and ways of working report up to 23 percentage points higher agentic AI readiness. They’re more likely to understand how to integrate agentic AI, feel motivated and confident in their skills, have opportunities to apply it, and perceive greater value in doing so. Interestingly, those feeling overwhelmed and struggling with the pace are often the ones facing these changes head-on – making them more ready to integrate AI agents, even when the process feels daunting. In fact, they’re 1.9x as likely to be high-frequency agentic AI users, showing that leaning into friction can accelerate adoption and impact.

These employees also report up to 18 percentage points higher realized individual value of agentic AI (RIVA). In other words, leaning into the challenge pays off: when employees avoid engaging with AI tools, they miss opportunities to build readiness and capture value. “Ignorance is bliss” may feel easier, but it delivers less impact.

This dynamic maps onto well‑established theory. Flow theory¹ suggests learning happens at the edge of our comfort zones; we need genuine challenge to grow. When employees encounter complexity and uncertainty, they are developing a new capacity to build relevant skills and knowledge. Mastery emerges when challenge meets skill, so without this friction, AI transformation might stall.

But not all friction and stress are created equal. Research shows there’s a difference between stressors that energize and signal opportunity versus those that drain and feel like obstacles. This distinction is called the Challenge–Hindrance Stressor Framework². When employees encounter friction with AI, they can interpret it as either type of stress – and that perception shapes whether they lean in or disengage.

A recent study³ applying this framework to AI found early signals that these interpretations matter:

• Mastery challenge stressors (e.g., learning new skills, tackling complex problems with AI) spark positive emotions, which boost adoption intention.
• Barrier hindrance stressors (e.g., technical breakdowns, unclear requirements) drive AI anxiety and reduce adoption intention.

If we want employees to move from overwhelmed to ready and confident, we need to shape the experience so agentic AI is perceived as a challenge (growth, mastery, value) rather than a hindrance (block, threat, anxiety).

Helping employees frame AI transformation as a mastery challenge

To better understand how to support employees facing friction during AI transformation, we dove deeper into the data and found three main levers that leaders and transformation change agents can pull: build trust, increase recognition, and provide training opportunities.

Build trust in the systems and the journey

Employees who are overwhelmed but still seeing value from AI agents are 1.7x as likely to trust AI agents than those who are overwhelmed and not seeing value. This higher level of trust may reduce employees’ tendency to categorize AI as a hindrance (e.g., fewer structural barriers, less anxiety). In the 2025 Agentic Teaming & Trust Report, we found that trust can be built by leadership role modeling, reliable systems, and peer advocacy. When we think about building trust in the face of challenge, consider:

• Encourage leaders and peers to share clear “why” and “how” stories of AI use that include failures and lean into continuous role modeling
• Invest in reliability and structure (e.g., consistent, well‑documented workflows and guardrails)
• Create peer advocacy opportunities, such as shadowing or live/recorded demonstrations, so employees see the “line of sight” from effort to outcome

Reliable systems and social proof can help shift employee appraisals from hindrance to challenge, increasing agentic AI adoption, readiness, and value.

Recognize and reward investing in development/upskilling

We found that employees who are overwhelmed but still seeing value from AI agents are also 1.7x as likely to report they are recognized or rewarded for using AI compared with those who are overwhelmed and not seeing value. For these employees, this level of recognition and incentives can increase positivity around experimenting, trying new things, and pushing themselves out of their comfort zone. For rewards and recognition, consider:

• Spotlighting success stories that celebrate learning and progress, not just perfection
• Rewarding employees who get involved in peer learning groups or pilot cohorts that test out new features and processes
• Tying recognition to experimentation, not just “getting it right” (e.g., “tried X, learned Y, improved Z”)

Recognition acts as a powerful signal, as it tells employees that effort matters as much as outcomes. When people see that experimenting and learning are valued, they’re more likely to embrace discomfort as part of growth rather than interpret it as risk.

Provide training that feels like a bridge, not a burden

Lastly, employees who are overwhelmed but are still seeing value from AI agents are 1.8x as likely to say they are adequately trained in how to use AI compared with those who are overwhelmed and not seeing value. Having training opportunities can support employees in understanding AI agents and their capabilities, which may turn the tide from an employee perceiving this transformation as a hindrance to a surmountable challenge. When considering training in the face of friction, consider:

• Delivering just‑in‑time, role‑specific learning modules/opportunities tied to real work outcomes (demonstrating the “why” and long-term ROI)
• Pairing training with coaching and on‑the‑job use, so the “learn” and “do” loops are closely intertwined
• Offering skill ladders/micro‑credentials that let people see and celebrate progress

Closing the AI skill gap is critical. Training helps employees feel supported as they stretch into new technology and ways of working. One research study found that 3 in 5 employees reported being more likely to use AI at work if proper training were available⁴ and technical self-efficacy was found to strengthen adoption pathways so increasing self-efficacy should also further push AI adoption and impact³.

Think about friction as fuel for growth, rather than failure. Organizations that frame AI as a challenge, not a threat, and reinforce trust, recognition, and training can turn moments of strain into opportunities for readiness and impact.

Stay tuned for our January Research Drop to keep up with what the Microsoft People Science team is learning! 

This month’s Research Drop analyzed 1,800 global employees from the Microsoft People Science Agentic Teaming & Trust Survey from July 2025. Employees were grouped in this analysis based on their favorability on two survey items (“I am overwhelmed by the idea of integrating agentic AI into my work” and “I am struggling to keep up with the pace of change surrounding agentic AI at work”) and composite scores on our Realized Individual Value of Agentic AI scale.

¹Csikszentmihalyi, M. (1990). Flow: The Psychology of Optimal Experience. New York: Harper & Row.

²Cavanaugh, M. A., Boswell, W. R., Roehling, M. V., & Boudreau, J. W. (2000). An empirical examination of self-reported work stress among US managers. Journal of Applied Psychology, 85(1), 65.

³Chang, P. C., Zhang, W., Cai, Q., & Guo, H. (2024). Does AI-driven technostress promote or hinder employees’ artificial intelligence adoption intention? A moderated mediation model of affective reactions and technical self-efficacy. Psychology Research and Behavior Management, 413-427.

⁴BBC. (May 18, 2025). Workers optimistic but overwhelmed by AI – study.