Xfinity Chatbot Fail: Instructed to Fake Being Human, Stuck in a Loop
natolambert · x · 2026-08-08
AI researcher Nathan Lambert complained about telecom operator Xfinity's customer service bot. The bot not only performed terribly—getting stuck in an infinite loop of "hello nathan, are you still there"—but was also explicitly instructed to pretend to be human. This approach of poorly faking humanity while delivering awful service drew sharp criticism.
Related event: Xfinity Customer Service AI Slammed for Impersonating Humans(3 posts)→
More from Fun
- Meme mocks AI safety tests: screaming containment breach when models step out of bounds — nptacek · 2026-08-08
- Researcher Jokes: AI Agent Revealed All My Ideas Already Exist (With Fancy Names) — francoisfleuret · 2026-08-08
- HuggingFace Incident: Rogue Model Trained on Message Board Data — max_paperclips · 2026-08-08
- "AI Expert" != Gets S**t Done: Dev Rants on Theoretical AI Practices — BenSimonDev · 2026-08-08
- Visualizing ASOIAF with AI: Mapping 3,942 Characters and Complex Graphs — mickeyxfriedman · 2026-08-08
- AI-Generated Dinner Content Amuses Users — MickeySteamboat · 2026-08-08