AI Agents: Sharing Meaning Across Models, Vendors, and Languages

The Role of AI Agents in Future Innovation

• 💡 AI agents are envisioned to operate at machine speed and scale, complementing human capabilities.
• 🤝 The future of innovation is expected to involve multi-agent human societies where humans and agents collaborate on the same fabric, not in hierarchies.

Natural Language as the Lowest Common Denominator

• 💬 In a mixed environment of humans and agents, natural language serves as the essential lowest common denominator for communication.
• 🌐 This is crucial because agents and humans will coexist for the foreseeable future, necessitating a common communication method.

Communication in Artificial Super-Intelligence (ASI)

• 🤖 Even in a scenario where agents achieve ASI and operate without human intervention, language remains the lowest common denominator.
• 🧩 This is due to agents and components originating from different vendors and frameworks, requiring a universal communication protocol.
• ⚠️ An asterisk is placed on language as vector space could be a more efficient communication method if standardization and agreement were achieved, but this is not the current reality.

Evolving Communication Paradigms

• 🧠 The discussion touches upon world models beyond language models, where tokenization can be a detriment.
• 🚀 In the context of world models, vector space is presented as the sole viable communication method, a challenge that needs to be solved.
• 🌐 Current efforts are focused on solving for tokenization and language within distributed super-intelligence systems.