---- On Thu, 12 Sep 2024 13:39:51 -0700 Daveed Benjamin wrote ----

Hi David,

Thank you for sharing your thoughts and insights—your experience at MIT’s Center for Collective Intelligence and the work with Tom Malone sound fascinating. The intersection of collective intelligence and AI agents is a timely and critical area, especially as we explore how to scale human collaboration effectively.

I agree that AI agents have immense potential in addressing the scaling challenges inherent in global networks. It’s particularly intriguing how they can help mitigate some of the cognitive limitations humans face when managing large-scale collectives. Your point about AI agents attentively listening to participants, rather than overwhelming us with information or ads, closely aligns with our vision for the Overweb. In the Overweb, we’re aiming to create an AI-assisted environment where humans engage meaningfully in a space that prioritizes privacy, data sovereignty, and fair value exchange, ensuring that AI enhances human capabilities rather than exploits attention.

This also brings to mind Engelbart protégé David Smith, whom I first encountered on a PCI call and covered in the Metaweb book. He described what he called the Augmented Conversation: a discussion between participants “extraordinarily enhanced” by computer tools and capabilities. In this model, the AI becomes a full participant in the conversation, allowing us to discuss and explore complex systems, datasets, and simulations as easily as we talk about the weather. A guarantee of shared truth is necessary: what I see, you must see. Any action taken that affects the shared simulation must reflect for all participants.

Without this, the communication channel is not trustworthy. The system must enable modification and extension of itself for all participants, allowing us to use it to extend its capabilities. Smith suggests a new kind of operating system built from scratch, focused on shared simulation and deep security.

I believe collective intelligence could greatly benefit from a shared information ecology, where related online information is connected at various levels of granularity, with explicit relationships (e.g., contradicting, supporting, citing). AI could assist by developing suggestions for these connections keeping a human-in-the-loop, while also relying on this global connected information network to improve the quality of its responses.

Your work with Jason Matheny on IARPA’s ACE activity resonates with my understanding of crowd wisdom. It seems like your research in emergent collaboration has strong synergies with what we’re proposing for the Meta-Layer Initiative. Specifically, we’re exploring how AI can support decentralized, real-time collaboration across interconnected, globally distributed networks, while ensuring that human agency remains central to decision-making.

I’d love to continue this conversation, especially around how these ideas could integrate with what we’re building in the Metaweb.



Warmly,
Daveed 

P.S. Our GPT-based AI, CyberSavant, had some thoughts on the projects you mentioned and suggested that the Overweb could significantly enhance and amplify their impact by enabling seamless, interconnected collaboration. You can see the prompt and response here: https://chatgpt.com/share/317e92eb-2965-4d5c-a302-e87c6ba611ff