Yonathan Cohen’s AI ‘Board of Directors’ Concept Explained

In a recent LinkedIn post, Yonathan Cohen discusses an innovative approach to leveraging artificial intelligence, likening it to creating an AI ‘board of directors’ to find and refine the best answers to complex questions. Cohen’s concept, inspired by Andrej Karpathy, utilizes multiple AI models to collaborate, debate, and ultimately deliver a superior, well-sourced response.

Cohen outlines a process where a question is posed, and several AI agents are tasked with searching the web and formulating answers. A crucial element of this system is the critical evaluation of each AI’s response by others. As Yonathan Cohen explains:

“They rank and criticize each other’s answers.”

This peer-review mechanism, Cohen suggests, is key to improving the quality and accuracy of the final output. He further elaborates on the final stage of his proposed system:

“A fifth AI picks the best answer with sources.”

The Mechanics of an AI ‘Board of Directors’

Yonathan Cohen details how this multi-AI approach functions, emphasizing the synergy between different models. The core idea is to harness the diverse capabilities of various AI systems, allowing them to act as a collective intelligence. Cohen notes the power of this collaborative method:

“Four AIs debating with real-time data. Best answer wins.”

Leveraging Real-Time Data for Superior Insights

A significant aspect of Cohen’s innovation is the integration of real-time data access. He highlights his own implementation, which uses Bright Data to provide each AI with unrestricted access to the full web. This ensures that the AIs are not limited by their training data but can pull current information to formulate their answers. In Cohen’s setup, specific AI models are employed:

• ChatGPT
• Perplexity
• Gemini
• Grok

By combining these powerful tools, Cohen aims to create a robust system capable of tackling a wide range of queries with unprecedented accuracy and depth. He provides a clear, actionable guide for others to experience this system firsthand.

How to Experience the AI ‘Board of Directors’

For those interested in testing this advanced AI collaboration, Yonathan Cohen provides a straightforward process. He outlines the steps required to interact with his AI ‘board of directors’:

1. Access the provided link.
2. Pose any question.
3. Receive the synthesized best answer derived from the collective AI input.

Cohen’s initiative demonstrates a forward-thinking application of AI, moving beyond single-model responses to a more dynamic, critical, and collaborative intelligence. This method promises to deliver highly reliable and well-supported answers, reflecting a significant step in the evolution of AI-driven information retrieval.