How Ethan Mollick's '5 & 5' Rule Transforms AI Into a Thinking Partner

Ethan Mollick, author of Co-Intelligence, has shared practical ways to use AI more effectively. His methods focus on refining prompts and pushing large language models (LLMs) to their limits. The goal is to turn AI into a tool that enhances thinking rather than just providing quick answers. Mollick recommends a '5 & 5' habit for complex tasks or decisions. This means refining a prompt at least five times and then generating up to 50 variations. By doing this, users can uncover unexpected but useful insights.

He also advises giving AI a clear character, context, and constraints. This approach increases the chances of getting high-quality responses. However, he warns against blindly copying AI output, stressing the need to verify facts and refine the results.

LLMs work by analysing vast amounts of text and predicting the next word in a sequence. While this allows them to map out detailed intellectual territories, they often produce false information—known as hallucinations—because they lack fact-checking abilities. Despite this flaw, Mollick believes relentless experimentation with AI can add a valuable layer of intelligence to work.

Mollick himself frequently uses ChatGPT to sharpen his own thinking. He suggests integrating one of the top four LLMs—ChatGPT, Gemini, Claude, or Grok—into daily tasks to maximise their potential. Mollick's techniques aim to make AI a more reliable partner in problem-solving. By refining prompts and verifying outputs, users can reduce errors and improve results. His book, Co-Intelligence, outlines these strategies but does not cover how major tech firms have adopted LLMs since its April 2024 release.