The Most Controversial Use of AI Yet: San Jose's Plan to Combat Homelessness

The city of San Jose, California is implementing a controversial new artificial intelligence system to detect homeless encampments throughout the city. The AI technology, developed by a startup called Zyrex, uses camera footage and machine learning models to identify tents, tarps, and other signs of outdoor living. The goal is to provide real-time data to city workers and homeless outreach teams to rapidly identify and address encampment sites.

Privacy advocates have raised concerns that the AI system is an invasive form of surveillance that could potentially criminalize homelessness. There are fears the technology could be expanded to track and monitor individual homeless people without their consent. The city maintains the program will simply allow more efficient allocation of resources and services to those in need.

San Jose has one of the highest rates of unsheltered homelessness in the country, with over 7,000 people living outdoors or in encampments as of 2022. City officials say traditional methods of reporting and responding to homeless camps have proven ineffective given the scale of the crisis. The $1.2 million AI pilot program is set to run for two years before being evaluated.

Zyrex claims its AI can detect camps with over 90% accuracy by analyzing patterns like clusters of tents and accumulations of debris and waste. However, experts warn that machine learning models trained primarily on data from whiter, wealthier areas may misidentify belongings of homeless people of color.

The ACLU has voiced opposition, stating "Systems of mass surveillance supercharge discrimination... We should be skeptical when governments want to automate operations that have historically driven criminalization."

(Source: "Revealed: a California city is training AI to spot homeless encampments", The Guardian, March 25, 2024)