New Orleans is testing Carbyne’s AI-powered Emergency Call Triage software
New Orleans is piloting Carbyne’s AI-powered system to triage 911 calls, particularly when many people phone in about the same incident so humans can focus on new or severe emergencies. Commenters are sharply divided: some see potential benefits in higher throughput, multilingual support, and handling surge events, while others fear misclassification, delays, and liability gaps when lives are at stake. Many argue that simpler automated messages or better funding for human dispatchers could address the same problems without introducing error‑prone, opaque AI into critical public safety infrastructure.
Scope and Intended Use of the System
- System is described as handling surges of calls tied to a single incident.
- AI agent asks if caller is calling about the known incident; if yes, it gives updates, if no, it forwards to a human.
- Some commenters think this is a narrow, reasonable use during major events (shootings, earthquakes, mass incidents).
Arguments in Favor of AI Triage
- Could reduce hold times when human dispatchers are overwhelmed or understaffed.
- A machine can handle hundreds of simultaneous routine/update calls, freeing humans for true emergencies.
- Potentially better language detection and translation than some human operators, helping callers with strong accents or other languages.
- Could quickly spin up incident-specific messages (“If you’re calling about X, press 1”), and provide standardized info.
- Seen by some as a pragmatic response to underfunded, overburdened 911 centers, not a replacement of humans.
Arguments Against / Risk Concerns
- Many see it as dystopian or premature, especially given LLM hallucinations and misinterpretations.
- Fear of misrouting, delays, and deaths in edge cases (children, panicked callers, unclear situations).
- Skepticism that AI support will stay “narrow”; concern it’s a wedge toward fully automated 911 and cost-cutting.
- Some argue this specific task doesn’t need AI at all; a simple IVR / prerecorded message + keypad input would be safer and more reliable.
- Worries about downtime, catastrophic failures, legal liability, and whether governments or vendors will be held accountable.
- Concern about bias and “predictive policing” if systems are extended beyond simple triage.
Governance, Incentives, and Accountability
- Several see this as a band-aid over chronic underfunding, chosen because “AI” budgets and political credit are easier than hiring humans.
- Fears of “accountability-washing”: failures will be blamed on the tech and vendor fixes, avoiding political responsibility.
- Some note governments may adopt risky AI because they can more easily dodge lawsuits than private firms.
Broader AI and Social Commentary
- Debate over whether AI tools generally make people dumber or simply reallocate cognitive effort.
- Comparisons to other automation (self-driving, calculators, physical labor) and historical Luddite debates.
- Side concerns about swatting, surveillance, and the company’s origins.
- Thread includes dark humor about failures and cultural references, reflecting deep unease.