Small idea: ML-aided search for citizen reporting

A sketch of tools for "What am I looking at? Who do I tell about it?"

Growth Stage: seedling

Created on

Last tended

Filed under technology

Just a little idea at the moment. I think one cool use of machine learning, both language and vision models, is to help citizens identify and log issues they see around the city.

So often when out in New York you'll see something that is awry—a bugged-out MTA train arrival screen, a burnt out light, etc—but you don't know who to tell or if they would care. A tool that could analyze your camera or video feed to assess what you're looking at, plot your location, find the relevant department to log it with, and submit a notification automatically would be very useful. Two major caveats:

  1. It would have to be very effective at identifying the issue and incorporating other input (text, voice) into a correct and coherent report to be not spammy to departments
  2. It would have to be completely opt-in by the user. I want no future where devices, software firms, or governments watch your camera and flag down "appropriate authorities" based on what they see without your express request.