New Orleans is putting artificial intelligence on the front lines of emergency response, and the experiment raises a question that cannot be answered with a marketing slogan:
When someone calls 911, how much room should there be for a machine to get it wrong?
The Orleans Parish Communication District, which handles more than 1,000 emergency calls a day, began routing some calls through an AI system in late July. The technology, called AI Emergency Call Triage, is designed to determine which callers need a human dispatcher immediately and which calls can be handled or filtered by the system.
Officials say the goal is straightforward: keep human dispatchers from being overwhelmed so they can concentrate on emergencies that demand immediate attention.
And there is a practical argument for that.
The system was first tested on New Orleans' non-emergency 311 line before moving into the 911 system. Officials say it is primarily being used when a caller is within roughly 200 meters of a crash that has already been reported or when every human call taker is already occupied.
Karl Fasold, executive director of the Orleans Parish Communication District, has pushed back against reports suggesting New Orleans is replacing its human dispatchers.
That is not what officials say is happening.
But the distinction does not eliminate the underlying concern.
A former 911 dispatcher identified online as Boxy, who says he spent 11 years working emergency calls, called the move “the worst idea you could possibly think of.”
His objection goes straight to the problem with trying to automate something that often depends on human judgment.
Some emergency calls are chaotic. Some callers cannot speak clearly. Some have heavy accents. Others may be terrified, injured or deliberately hiding what is happening.
A domestic-abuse victim, for example, might pretend to be ordering a pizza rather than openly telling a dispatcher that someone is threatening her.
In situations like that, a human dispatcher isn't merely processing words.
The dispatcher is listening for what the caller is not saying.
“You need a person,” Boxy wrote.
That warning deserves more than a shrug.
New Orleans appears to be the first major American city to allow AI to actually answer some emergency calls. Other jurisdictions have experimented with narrower applications.
Atlanta, for example, has used AI to help human dispatchers determine caller locations and retrieve relevant medical information. Seattle began using similar technology in 2024 to help triage certain medical calls.
The potential benefit is obvious. A properly designed system could reduce unnecessary workloads, identify duplicate calls and help human dispatchers get to the most urgent emergencies faster.
But the risks are just as obvious.
AI systems can generate convincing information that is simply wrong. Voice systems face an additional challenge because emergency callers are rarely speaking calmly, clearly and under ideal conditions.
There is also the cybersecurity problem.
Researchers have demonstrated that voice models can be manipulated, raising another uncomfortable question for emergency services: What happens when the technology itself becomes the target?
Officials in New Orleans insist that AI can be used safely with human oversight, careful training, strong cybersecurity protections and strict rules governing what the system is permitted to do.
That is the right standard.
Technology should serve emergency personnel — not become an excuse to push human judgment out of the room.
There is nothing inherently wrong with using AI to help first responders. If a machine can identify duplicate crash reports or relieve an overloaded call center while immediately escalating genuine emergencies to trained professionals, that could save time and potentially save lives.
But 911 is not a customer-service chatbot.
The person on the other end of that line might be trapped in a burning house, hiding from an attacker, suffering a heart attack or whispering because speaking normally could put someone in danger.
When seconds matter, Americans should never be told that a machine was “mostly right.”
New Orleans has opened the door to a potentially valuable tool. Now officials have the responsibility to make sure that door never becomes a replacement for the human judgment that emergency calls have depended on for generations.
