In this video, Shane Fozard introduces the concept of Tree of Thoughts Prompting, which was shown by Stanford University to produce 900% better results with AI models compared to single-shot prompting. Shane demonstrates how Tree of Thoughts Prompting allows AI to argue with itself, find gaps, and iterate its responses to deliver a more refined and comprehensive final answer.
To make this process accessible, Shane has built a GPT called Tree of Thoughts 3 Experts, which automates the Tree of Thoughts Prompting method. This GPT selects three relevant experts to discuss and refine the input question, ensuring that the final output is well-considered and improved through multiple iterations.
Shane shows how to use the Tree of Thoughts 3 Experts GPT by asking it various questions. For example, he asks, “What is the best structure for a social media post?” The GPT selects experts like Seth Godin, Nancy Duarte, and Robert Cialdini to provide insights on different aspects of social media posts. Each expert offers their perspective, and the GPT synthesizes these into a comprehensive recommendation.
He further demonstrates the GPT’s capabilities by asking it to refine the structure into actionable steps and even to generate a story prompt. The GPT iterates through the experts’ feedback to enhance the prompts and provide more detailed and effective outputs.
Shane emphasizes that this approach helps users get beyond the initial, often suboptimal response from AI, and instead achieve a more polished and thoughtful result. He encourages viewers to try the Tree of Thoughts 3 Experts GPT for various use cases, from marketing strategies to creative writing, to ensure they are getting the most out of AI.
Overall, the video highlights the power of iterative prompting and the benefits of using a structured approach like Tree of Thoughts to enhance AI-generated content.