AI adoption numbers are skyrocketing. Boston Consulting Group reports that 74% of front-line employees use AI at least two to three times a week, up from 51% in BCG’s 2025 report.
AI champion programs have played a significant role in driving that growth. By deploying trained internal employees to provide on-the-ground encouragement and training to their peers, organizations are turning skeptics into daily users. Citi’s usage of its internal AI tool has increased from single digits at the time of its initial rollout to more than 80% of its workforce adopting it, and law firm Ropes & Gray increased its use of AI from a few hundred prompts per month to more than 282,000 across 2,200 employees.
But organizations are pushing AI adoption faster than they’re building the guardrails to use it well. One recent survey found that more than half (51%) of employees are receiving conflicting guidance on when and how to use AI at work. Of those using the tools, only 7.5% of employees have received extensive AI training. Another 23% have received no training at all.
Driving adoption is the easy part. Things get more complicated when AI use moves beyond productivity tasks into team workflows, and eventually into high-stakes people decisions. Each step shifts team dynamics, and without meaningful context, higher AI adoption makes it easier for teams to approach those moments with incomplete information.
What happens when AI adoption outpaces our understanding of its impact on people?
AI adoption gets tracked the same way we track any other product rollout, through usage rates, prompt volume and time saved. On those metrics, AI is proving its value. But those numbers miss the only question that actually matters for a people leader: Is AI making my managers better at managing?
Answering that requires a different set of measures, such as tracking whether:
- Feedback conversations land better
- Conflicts reach a resolution
- Employees feel understood
- Turnover on a manager’s team improves after AI enters the picture
Most organizations aren’t tracking any of that. The tools keep spreading without guidance on how to use them for leadership development.
Instead, organizations are effectively leaving managers to figure out what appropriate use of the tool looks like, without the training or context to make those decisions well. PI’s own Amy Saenger watched it happen on her own team. Her marketers were three tools deep in experimentation before she’d even settled on a rollout plan.
“My team didn’t need permission to move fast. What they needed was someone willing to build the guardrails after the fact instead of before.”
Amy Saenger, Director, Demand Generation Team @ PI
Missing guardrails is only part of the problem. Even with them in place, a manager still needs to know the person they’re talking to. A manager might turn to an AI tool to deliver feedback, resolve a conflict or carry out a discussion about someone’s future. But that tool knows nothing about the person sitting at the other end of the table.
Compounding the problem, nearly half of managers have received no formal management training. That means organizations are chasing adoption numbers before they’ve established the training or managerial support needed to use it well. When that happens, AI can scale inconsistency just as easily as it scales productivity.
Where should AI stop and human judgment begin?
AI earns its keep on plenty of managerial tasks. It can draft a first pass at a difficult email, summarize a meeting nobody had time to fully absorb or organize a set of talking points before a one-on-one. Those are process problems, and AI is a genuinely useful process tool.
But when it’s time to give some tough feedback or manage conflict between two teammates, a generic AI tool can only do so much. It can’t tell the manager whether the employee they’re about to speak to needs time to sit with feedback or wants it delivered straight. And neither can it explain an individual’s behavioral style, like what makes them tick or where they’re likely to get defensive and shut down.
No matter how good the prompt is, if AI tools lack human context, managers can’t engineer their way through a tough conversation. If managers are going to use AI to support their people, they need tools that accurately understand the people they’re being asked to lead.
What organizations should do in the meantime
You don’t have to wait for a perfect strategy to make AI more useful for people decisions. Here are a few practical steps to get you started.
- Draft a short policy on when AI is and isn’t appropriate for feedback, conflict or performance conversations to provide users with a baseline.
- Feed AI context about the person, specifics on how an employee communicates, processes feedback and responds under pressure, so the guidance it gives is actually useful.
- Loop in HR on data sources like behavioral assessments, performance history and engagement data that already exist in most organizations, and see whether any of it can connect to your AI tools to support more accurate output.
But even the best prompts and policies only go so far without a validated, science-backed foundation for understanding your people.
How can AI give managers better insight into their people?
PI’s Behavioral Assessment provides that foundation. Grounded in more than 70 years of behavioral science, the assessment gives managers a common language for understanding the people on their teams. When those insights are brought into the AI tools managers use every day, they become easier to access and apply when it matters.
Instead of prompting AI with “How should I give this employee tough feedback?”, a manager can pull behavioral data in moments and use that context to shape the conversation. With a validated tool, they know whether someone needs space to process or wants directness. They know what actually drives that person and where friction is likely to show up before it becomes a problem.
Adoption numbers tell you how many people are using the tool. They don’t tell you whether the people using it are getting better at leading. Behavioral science gives managers the context to turn AI from a tool that tracks usage into one that actually helps them lead.
See how PI’s Behavioral Assessment and AI-powered tools combine behavioral science with AI to give managers deeper insight into their people and put that insight to work when it matters most.