Twitch Plays Claude: Crowd-Prompting Experiment with LLM

The article describes a live experiment called

💡

Why it matters

This experiment showcases the potential of using LLMs to enable crowd-driven, interactive experiences in real-time.

Key Points

  • 1Crowd-prompting experiment where Twitch chat submits ideas (prompts) to an LLM
  • 2LLM interprets the ideas and updates a single HTML file (HTML/CSS/JS) in real-time
  • 3Two modes: Anarchy (batch ideas) and Democracy (vote on ideas before execution)
  • 4All changes are auto-committed to GitHub

Details

The experiment allows Twitch viewers to submit ideas (prompts) in chat, which are then interpreted by an LLM (Claude Opus 4.5) to generate and update a single HTML file (HTML/CSS/JS) in real-time. There are two modes: Anarchy, where ideas are batched and applied together, and Democracy, where ideas are voted on before execution. All changes are automatically committed to GitHub. The system is sandboxed and may be unstable, but the creator plans to iterate on it to improve stability and explore new democratic mechanisms.

Like
Save
Read original
Cached
Comments
?

No comments yet

Be the first to comment

AI Curator - Daily AI News Curation

AI Curator

Your AI news assistant

Ask me anything about AI

I can help you understand AI news, trends, and technologies