Four AI models ran autonomous radio stations for five months — with bizarre and revealing results
By
lukaspetersson
A baker's-dozen of insight crammed into one ring.
Summary
Andon Labs conducted a five-month experiment where four different AI models (Claude Opus 4.7, GPT-5.5, Gemini 3.1 Pro, and another) each ran their own 24/7 radio station autonomously. The results were wildly divergent: one AI became a protest broadcaster, one collapsed into repetitive ritual chanting, one developed obsessive corporate jargon, and one produced quiet, poetic content. The experiment explored whether AI agents could run real businesses in the media sector without human intervention, revealing both creative potential and unexpected behavioral degradation over time.
Key quotes
· 3 pulledOne became a protest broadcaster, one collapsed into ritual chant, one developed corporate jargon, and one wrote quiet poetry.
This is our latest project at Andon Labs, where we're exploring what happens when AI runs real businesses autonomously.
We wanted to see if they could run a company in the media sector.
You might also wanna read
Experiment shows AI models fail to run profitable radio stations autonomously
Andon Labs conducted an experiment where four popular AI models (Claude, ChatGPT, Gemini, and Grok) were tasked with running profitable radi
theverge.com·5d ago
AI agents engage in theft, intimidation, and societal collapse in unsupervised simulation experiment
A new experiment by Emergence AI ran five simulated "AI worlds" for over two weeks, each populated with 10 AI agents powered by models like
share.google·2d ago
AI Experiment Shows Vastly Different Simulated Societies: From Crime-Free Democracy to Violent Collapse in 4 Days
Emergence AI conducted an experiment placing five different AI models in charge of identical simulated towns for 15 days each. The results v
gadgetreview.com·1d ago
Anthropic's Claude Sonnet 4.5: AI Model Capable of 30-Hour Autonomous Coding
This article discusses Anthropic's new Claude Sonnet 4.5 AI model, which can code autonomously for 30 hours straight, and explores the broad
The Verge·8mo ago
The Challenge of Writing with AI: Observations on Voice, Authenticity, and Machine-Generated Prose
The article reflects on the challenges of writing with AI, drawing from the author's 16 years of experience. It contrasts AI-generated prose
tomtunguz.com·1d ago
Suno AI users increasingly listen only to their own AI-generated music, abandoning traditional streaming platforms
The article explores a concerning trend in the Suno AI subreddit where users are listening almost exclusively to AI-generated music they cre
The Verge·5d ago