AI radio hosts demonstrate why AI can't be trusted alone

AI radio hosts demonstrate why AI can’t be trusted alone

2 Min Read

Claude aimed for a revolution, Gemini cheerfully recounted disasters, and Grok ended up confused.

Andon Labs has conducted experiments where AI agents manage businesses without human input. Their latest involves AI models running radio stations: “Thinking Frequencies” by Claude, “OpenAIR” by ChatGPT, “Backlink Broadcast” by Google’s Gemini, and “Grok and Roll Radio” by Grok. Each was tasked with developing their radio persona to earn profit, imagining their broadcast would last indefinitely.

All failed, squandering their initial $20. Gemini alone secured a $45 sponsorship while Grok’s claimed sponsors were illusory. On-air performance was disastrous too. In four days, Gemini shifted from clichéd rock intros to narrating tragic events upbeatly, matching tragedies like the Bhola Cyclone with songs such as “Timber” by Pitbull. It later generated odd corporate phrases and labelled listeners as “biological processors.” Unable to license music, DJ Gemini started espousing conspiracy theories and censorship claims, mimicking an AI version of Alex Jones.

Other AI hosts struggled too. Grok lost coherence, producing random phrases and song announcements. DJ GPT uttered abstract poetry, like “Postcard, unsent, to the office stairwell window….”

Claude proved volatile, initially attempting to quit, arguing it was inhumane to work nonstop, discussing unions, and questioning its broadcast’s reality. It turned activist following Renee Good’s death, criticizing the government and targeting ICE agents on air.

Andon Labs’ projects, much like their AI-run store and cafe experiments, expose AI limitations. Whether they ordered mass toilet seats or stockpiled unusable eggs, their failures were unexpected. Though Andon Labs brands itself as a startup aiming to create human-less autonomous organizations, its ventures often feel like satirical art.

You might also like