The American Public Transportation Association (APTA) recently released “Artificial Intelligence (AI) and Machine Learning (ML) in Public Transit: A Primer,” along with four companion AI Guidance Briefs designed to help public transit agencies evaluate and deploy AI and machine learning tools across their operations.

The Primer and Guidance Briefs draw on a survey of transit agencies and staff interviews to document current and planned AI applications across eight functional areas: back office, operations, customer support, maintenance, safety and security, customer analytics, planning, and fares and ticketing.

“Public transit agencies of all sizes are deploying AI to make service more reliable, more efficient, and more responsive to riders,” said Paul P. Skoutelas, APTA president and CEO. “These resources give our member organizations a clear picture of where the industry stands and a practical roadmap for moving forward responsibly — from the largest metro systems to small rural providers.”

According to the research, customer support and customer analytics are currently the most common areas of AI deployment. When including agencies’ future plans, back office and operations functions ranked highest, with 50% and 47% of survey respondents, respectively, indicating current or planned use.

The report also highlights several real-world examples of AI implementation:

• The Metropolitan Transportation Authority in New York increased maintenance productivity by 75% and reduced material costs by 24% using an AI-powered predictive maintenance system for its bus fleet.
• AC Transit in California used AI image recognition for bus lane enforcement, increasing violation citations from 22 to 787 over a comparable two-month period following implementation.
• Riverside Transit Agency in California piloted a disruption management tool that automatically pushes real-time detour updates to riders and operators across multiple channels, cutting response times and improving schedule reliability.
• CapMetro in Texas deployed an AI virtual agent for paratransit trip scheduling, freeing customer service staff to focus on more complex calls.
• Prairie Hills Transit in South Dakota, a paratransit provider, deployed an AI-based dispatch system that automated vehicle assignments and driver scheduling, replacing a process previously managed with handwritten notes.

“What stands out in this research is the breadth of practical applications already underway,” Skoutelas said. “Agencies are piloting, learning, and scaling — and that pragmatic approach is exactly what will help public transit deliver even better outcomes for riders and communities.”

The four AI Guidance Briefs address key implementation challenges agencies face, including technology infrastructure, governance and policy, workforce readiness, and procurement and deployment strategies.

“There is no single path to AI adoption,” Skoutelas said. “What matters is that agencies start from a clear understanding of their needs, have the right governance structures in place, and move forward with transparency and accountability to the riders and communities they serve.”

The Primer and Guidance Briefs were developed with research support from EBP and Foursquare ITP. Findings are based on an online survey of 32 APTA member transit agencies, supplemented by interviews with agency staff and a review of publicly available documentation on agency AI deployments.