Particle.news
Download on the App Store

San Francisco Billboard’s Fake ‘AI’ Chat Goes Viral and Returns With Human Moderators

Bryant has relaunched ChatTJB with vetted volunteers, moderation filters and crisis resources to manage a surge of submissions and test whether a human-run chat can scale.

Overview

  • The project debuted after Tucker James Bryant put a SoMa billboard live on July 27 that billed ChatTJB as an AI chat while a small asterisk explained that “AI” means “average individual.”
  • After a viral Instagram post the site flooded with traffic, peaking at about 5,000 queries in an hour and accumulating roughly 30,000 total submissions across days.
  • Bryant temporarily paused the service to avoid burnout and now has reopened it with a small roster of vetted volunteer responders, an onboarding quiz, message triage and filters for abusive content.
  • Bryant used AI tools to build the site but says every reply is written by a person, sometimes aided by web searches, and the project includes an optional low-cost support tier that few people have paid.
  • As an art stunt critiquing blind trust in machine outputs, ChatTJB highlights limits of automated-style interfaces, raises questions about user safety and privacy when humans read private queries, and shows how physical ads can drive large online attention.