James Evans develops the idea that intelligence is not a monolithic quantity, but a plural, social transition – a combinatorial complexification of billions of humans and non-biological agents.

As large AI models are increasingly built by other large models, we see the automatic emergence of recursive complexity. Frontier reasoning models, tuned by reinforcement learning, succeed not by “thinking longer” but by simulating internal “societies of thought” – recursive debates among distinct cognitive perspectives and personalities that causally drive reasoning accuracy.

This reflects the wisdom of human crowds in science and society, which succeed as a function of distinct perspectives on complex problems. He also shows that when perspective complexity is suppressed within machines, humans, or human-machine systems, performance across a range of challenges collapses.

James Evans proposes the importance of evolving complex social institutions that facilitate the continuous evolution of relevant diversity to facilitate the AI-human intelligence explosion.

Please note that we are at full capacity & registration is closed.