The Montgomery County Emergency Communications Center (ECC) may start using artificial intelligence (AI) to handle non-emergency 911 calls by the end of 2026, according to a council public safety committee on Oct. 6. The proposed system would screen calls, provide information to callers directly or re-route them to the appropriate department, all without human intervention.
Currently, the ECC answers 75%-85% of calls within 15 seconds and 13%-23% of calls after 20 seconds. Between 4 and 5 p.m., one of the busiest hours of the day, wait times sometimes exceed 35 seconds. Research shows a year-by-year increase in response times, starting in 2022, when 15% of calls went to an automated queue and the average hold time was seven seconds. In 2024, 24% of calls went to the automated queue and the average hold time was 11 seconds.
During the committee meeting, ECC Director Jennifer Reidy-Hall said that the department aims to address these numbers and get emergency callers the help they need. Non-emergency calls make up 40-50% of total 911 traffic to Montgomery County, she said.
“We hope that [AI], paired with increased staffing and implementing more efficient schedules for our specialists, will move us closer to the goal of being able to answer 911 calls with zero wait time,” Reidy-Hall said.
In August, Fairfax County tested an AI response system on its non-emergency line. The system handled 150 calls for about two hours and was able to communicate in English and Spanish, mainly re-routing callers to the appropriate county agency.
Counties such as Snohomish County, Wash., Arlington County, Va. and Hamilton County, Tenn. have piloted similar technology for emergency lines. Monterey County, Calif., reported a 30% reduction in 911 calls requiring human interaction and a 7%-10% increase in overall operational efficiency in April 2024 after implementing call-taking AI.
Junior Siena Calderon said she was concerned about the safety of using AI over humans, but otherwise supported the potential change.
“It makes a lot of sense to do this,” Calderon said. “My only worry would be if the AI doesn’t correctly identify if it’s an emergency or not.”
Dr. Anton Dahbura is an associate research professor at Johns Hopkins University in its Department of Computer Science and a member of its Data Science and AI Institute. Dahbura said AI is constantly available, which reduces dispatcher overload and allows humans to focus on critical situations. However, he warned about potential limitations.
“Current AI systems aren’t great at realizing when they don’t understand something,” Dahbura said. “That’s why there must always be an easy way for human dispatchers to intervene.”
In addition to using AI, the ECC plans to reduce hold times by filling employee vacancies and adjusting staff shifts. Reidy-Hall said that the number of operators is even throughout the day, but call volume is not. In response, the center will schedule more dispatchers to work during peak hours, such as 1-2 p.m.
Public Safety Committee Chair Sidney Katz also suggested using AI when a large number of calls report the same emergency, like a vehicle crash. The system could inform callers from the location that the incident had already been reported and ask if they have any new information.
In 1950, Alan Turing’s “Turing Test” first explored a machine’s ability to exhibit human-like intelligence. Now, AI is common in everyday life: safety features in cars, generative chatbots and social media algorithms. Dahbura said the new 911 systems prove how far AI has come.
“Machines can now carry on conversations, classify information and act on it in real time,” Dahbura said. “The technology is powerful, but how we design and monitor it, especially ensuring that humans stay involved, will determine whether it’s truly safe and effective.”
