Transit agencies investing in AI often find that procurement, implementation, and configuration go smoothly enough. What takes longer, and what decides whether the tool actually gets used, is earning the trust of the people being asked to work alongside it.

This isn't a technology problem. It's a change management problem, and it deserves as much planning as the tool selection itself.

Why staff skepticism makes sense

It would be easy to frame resistance to AI as technophobia, but that misses what's actually happening.

Dispatchers and operations staff have built their expertise over years. They know the quirks of their service area, the regulars on their routes, the edge cases that don't show up in any manual. When an AI system arrives with confident recommendations, the instinct to question it isn't irrational. It's professional.

They're also watching broader narratives about AI and employment. Whether or not automation is actually threatening their specific role, the ambient uncertainty is real. Agencies that don't address it directly leave that uncertainty to fester.

When Pinellas Suncoast Transit Authority (PSTA) launched Spare AI Voice for its paratransit and on demand services, the goal was simple. Let riders book, cancel, and check trip status without waiting on hold. The implementation also sent a clear internal signal that the agency was investing in tools that reduced workload for call center staff, not replacing them.

PSTA was upfront about what the system was for from the start, publishing a press release explaining exactly how AI Voice worked and what riders and staff could expect. Today, roughly one in four PSTA paratransit bookings is handled by AI Voice.  Staff who were fielding every call now have more capacity for the complex scheduling and exception-handling work that still requires a human.

The transition worked because the agency was direct about what the tool was and wasn't meant to do, before it ever went live.

Start with transparency about what the tool does

The first step to building trust is making the system legible. Staff should understand, in plain language, what the AI is doing, what data it's working from, and where its authority ends.

If an AI tool is handling booking calls, dispatchers should know how it decides to escalate to a human. If it's making trip assignment suggestions, they should understand the logic behind those suggestions and know they can override them. If it's flagging exceptions, they should know the thresholds it uses.

Opacity breeds distrust. Transparency, even when the system isn't perfect, creates the conditions for a working relationship between staff and tool.

This applies at the leadership level too. Texas DOT's CIO, Anh Selissen, has been public about the fact that her agency evaluates every AI tool against clear standards before it touches agency data, including encryption, access controls, and privacy governance. That kind of visible accountability reassures staff that someone is paying attention to the risks, not just the upside.

Involve staff before launch, not after

One of the most consistent mistakes agencies make is presenting staff with a finished product. By the time a tool is live, the decisions about how it works have already been made. Staff are left to adapt to a system they had no hand in shaping.

Involving frontline staff earlier changes this dynamic. Pilot groups who test tools before launch become advocates rather than skeptics. Staff who flag an edge case during testing feel heard. Supervisors who help define escalation logic become invested in the system's success.

This doesn't require a lengthy process. Even a few structured sessions with dispatchers and drivers, asking what they find most tedious, what they'd want flagged automatically, and what they'd never want the system to decide alone, will meaningfully improve adoption and surface requirements your vendor may have missed.

Moventis, a transit operator in Spain, took a similar approach when rolling out automation across its operations. Rather than deploying AI tools and hoping for buy-in, the team pursued a broad digital transformation strategy across its operations. This includes digitizing previously manual processes and implementing integrated systems that connect fleet, operations, and commercial functions, to improve planning, efficiency, and the development of new mobility models.

Treat early failures as learning, not evidence

No AI tool performs perfectly from day one. How an agency responds to early errors is one of the most important trust signals staff receive.

If a dispatcher overrides an AI recommendation and turns out to be right, that should be acknowledged, not treated as a failure. If the tool makes a mistake that creates extra work, staff should have a clear channel to report it and see it addressed.

The goal isn't to defend the technology. It's to demonstrate that leadership is paying attention, that feedback is taken seriously, and that the system is being improved based on real-world use.

Agencies that build this feedback loop tend to see trust grow over time. Staff who feel their experience matters are more likely to use the tool consistently, which generates the data needed to make it better.

Reframe who the tool is working for

The framing that lands best with operations staff isn't "AI is the future." It's "this tool handles the tedious parts so you can focus on the work you're great at."

Dispatchers don't go into transit because they enjoy logging cancellations. They go into it because they excel at managing complexity under pressure, building relationships with riders, and solving problems in real time. AI that takes the tedious, manual tasks off their plate and creates space for more of the work they enjoy.

When agencies lead with this message, and back it up with tools that deliver on the promise, skepticism tends to soften. Staff who felt threatened by the technology start to see it as something they own, not something that owns them.

Trust isn't built through a launch announcement. It's built through consistency, clear communication, genuine involvement, honest responses to problems, and evidence over time that the system is working in their interest as much as the agency's.