AI has become a regular part of daily life since ChatGPT made generative models widely known in 2022. People use it to draft emails, edit photos, find entertainment recommendations, and automate routine tasks. Another quieter application has emerged in public safety: an AI system now assists with a portion of 911 calls in New Orleans. Orleans Parish officials confirmed this week that artificial intelligence is being used to handle a small, specific category of emergency calls in the city’s dispatch network.
What the AI actually does
The AI system is limited to a narrow, well-defined function. It activates when a 911 call originates near a motor vehicle crash that has already been logged in the city’s dispatch system, specifically within roughly 200 meters of the recorded incident. When the system determines the caller is reporting the same crash, it informs them that emergency responders are already en route. The automated assistant then asks callers to remain on the line only if they were directly involved in the crash or if they have new, relevant information. Callers who are merely bystanders or who are confirming information can be advised they may hang up.
According to Orleans Parish Communications District director Karl Fasold, the tool addresses a practical problem: during rush hour, a single crash can generate dozens of duplicate calls that flood the phone lines and delay response to other emergencies. By filtering repeat reports of the same incident, the AI helps prioritise human resources and reduce congestion in the call queue. The district reports that the automated system now resolves about 15 to 18 out of every 20 repeat crash-related calls of the type it is designed to handle.
Officials emphasize that the AI is not used for violent crimes or life-threatening situations. For calls that signal imminent danger, human dispatchers remain the first responders on the line. The automated assistant is deliberately constrained so that it only intervenes where the risk of misclassification is low and the consequences of an error are limited.
Why some experts want more guardrails
Not everyone is fully comfortable with deploying AI in emergency communications. Researchers and professors who study the use of AI in public services have raised concerns about transparency and oversight. One issue is that there are currently no widely adopted city or national policies that specifically govern how AI should be used in emergency response settings, including requirements for disclosure to callers. In the New Orleans deployment, callers are not always explicitly informed they are interacting with an automated system rather than a human operator, which some experts say raises ethical and practical questions.
Academic observers caution that even a single significant error in triaging emergency calls could have serious consequences. As Rob Lalka, a researcher who has commented on AI in public services, has noted, mistakes in high-stakes contexts can put lives at risk. These concerns have led some observers to call for clearer rules, independent audits, and robust testing before expanding similar tools to other lines.
New Orleans is reportedly preparing to adopt a similar AI-supported approach for its non-emergency 311 line, which would use automation to handle routine inquiries and reduce the load on human staff. If expanded, such systems will likely be subject to continued scrutiny from policy makers, privacy advocates, and public-safety professionals who want to ensure that automation improves service without undermining public trust or safety.
As municipalities experiment with automated assistants in dispatch centers, the balance between efficiency gains and the need for safeguards remains central. Clear policies on disclosure, limits on automated decision-making, oversight mechanisms, and regular evaluation can help communities make informed choices about integrating AI into critical public services while minimising risks.