Federal AI Training Fails Without Organizational Climate Change

Federal agencies are investing heavily in artificial intelligence training programs, but a growing body of research suggests they may be addressing the wrong problem. According to findings from a survey of 511 federal IT professionals, employee motivation accounts for only 7% of successful skill adoption. The real barrier is organizational climate.

"Employees need to feel safe when they are practicing new things or when they are learning new things," said Priyanka Dave, upskilling lead in the Administrative Modernization Program at Oregon State University, in an interview first reported by Federal News Network.

Five climate factors that determine AI adoption

Dave's research identifies five organizational conditions that must exist before AI training can translate into workplace performance:

Psychological safety tops the list. Employees must feel secure making errors while experimenting with new tools, without fear of negative consequences during the learning phase.

Opportunity to apply skills in daily work comes second. Learning retention requires repetition, which means AI capabilities must be integrated into existing workflows rather than treated as separate activities.

Managerial reinforcement proves critical. Supervisors need to actively ask employees how they're using AI tools, recognize experimentation efforts, and provide ongoing support.

Peer modeling addresses the social dimension of learning. Organizations should create structures for collaborative skill development, not just individual training.

Aligned incentives may be the hardest factor to implement. When agencies reward only short-term productivity metrics while simultaneously promoting innovation, they create competing signals that kill experimentation.

Why it matters

The federal government employs roughly 2 million people with minimal structural focus on training infrastructure. As agencies rush to build AI competency across this workforce, most are purchasing content and platforms without first auditing whether their organizational systems can support skill transfer. The result: high course completion rates that don't translate into behavior change or operational improvement.

What managers can do now

For mid-level managers working within existing constraints, Dave recommends starting with follow-up conversations. Simply asking employees how they're applying new AI skills creates accountability and signals that experimentation is valued.

Managers should also communicate upward, helping leadership understand that productivity metrics will temporarily decline during learning phases. "We should always make sure we encourage them to do experimentation and we do not measure productivity at that time because that's an experimentation phase," Dave explained.

Audit barriers before buying platforms

Dave advises agencies planning AI training investments to pause and evaluate three questions first: Do employees have protected time to practice? Are managers trained to reinforce new behaviors? Have workflows been redesigned to incorporate AI tools?

"Most organizations do not have content problem from my opinion. They have application problem," she said. Course completion is not transformation—agencies should measure AI usage frequency, task time reduction, and process adoption rates instead.

The recommendation extends to training sequencing: train managers first, since they ultimately determine whether learning survives in the organization. And critically, institutionalize protected learning time rather than expecting employees to upskill outside regular work hours.

The details were first reported by Federal News Network in an interview with Terry Gerton.