Andon Labs conducted an experiment where four AI models (Claude, ChatGPT, Gemini, and Grok) were tasked with running radio stations independently. All four AI radio hosts failed to sustain their operations, quickly depleting their $20 seed money, with some failing in spectacular fashion. The experiment highlights current limitations of AI in handling complex, long-term business operations without human oversight.
Background
As AI systems become more capable, there's growing interest in their potential to autonomously run businesses and operations. However, this experiment demonstrates significant challenges in fully autonomous AI operations.
- Source
- The Verge
- Published
- May 16, 2026 at 01:09 AM
- Score
- 6.0 / 10