Many organizations invest in AI tools, train employees to use them, and expect adoption to follow. But this isn’t always the case. A similar pattern repeats across global workforces: AI adoption falls short of full usage, and companies fail to realize the full potential of their investments. One Fortune 500 HR company experienced this firsthand. During AI deployment, the company provided employees with training and clear internal communications, but adoption still stalled at 58%. Addressing obvious barriers such as accessibility and skill wasn’t enough on its own. Many of the hidden barriers blocking adoption, it turned out, were behavioral.
The trust gap
The company approached our team at The Decision Lab to design an intervention to address their adoption problem. After conducting an extensive literature review and mapping employee adoption journeys via workforce interviews, surveys, and expert consultations, trust emerged as a key barrier. Uncertainty around the perceived complexity of AI tools and low trust in AI accuracy were among the main behavioral barriers stalling adoption.
At the same time, we discovered that employees fell into two distinct trust profiles: enthusiasts, who were actively optimistic about the potential of AI, and skeptics, who feared job disruption and remained doubtful about the value of AI in the workforce. These two profiles diverged sharply in how they integrated AI into their workflows and championed broad-scale adoption. Enthusiasts actively sought out and engaged with advanced tools while serving as advocates for AI within the organization. Skeptics, on the other hand, were easily frustrated by AI’s flaws, felt disengaged with the technology, and used it only when absolutely necessary. While enthusiasts were eager to work through AI’s limits, skeptics were more easily stalled by poor integration and felt unsupported when they ran into issues.
Designing for confidence
Based on these findings, we developed a custom framework to address behavior barriers to adoption, specifically tailored to the company’s workflows. Through an iterative design process, we ran pilots with representative teams and incorporated feedback from employees, team leaders, and behavioral science specialists to ensure our interventions were both practical and evidence-based. We landed on eight targeted interventions, all of which were tested for one month with over 100 employees. During the pilot, we tracked data on employee confidence and ease of using AI to determine whether our interventions were actually moving the needle on the trust gap.
Following the pilot, employees reported a 41% increase in their confidence exploring AI tools and a 220% increase in ease of use. Overall, the pilot resulted in a 35% increase in organization-wide AI adoption, up from the pre-pilot baseline of 58%.
Built to scale
These interventions were deliberately designed and selected to be easy to implement and scale. Each came packaged with simple handouts and guides so teams could seamlessly integrate them into existing workflows without requiring time-intensive facilitation or specialist support. That simplicity is key for an organization of this size: the intervention is now ready to scale across the company’s 20,000-person workforce. Our framework also included mechanisms to adapt to teams’ specific barriers and adoption maturity, such as a diagnostic that allowed teams to select the most relevant interventions.
The implications aren’t specific to a single organization. Technical barriers, such as issues with access, employee skill, and integration, tend to be the most visible AI adoption challenges and the easiest to solve. As such, they’re often the issues organizations reach for first. But for large, distributed workforces, these barriers rarely represent the full picture. Behavioral factors, such as people’s willingness to experiment with AI and trust the value of these tools, are ultimately what decide whether adoption sticks.