"Artificial Symbiotic Intelligence" proposed as alternative to the singularity

The essay extends an argument the authors' circle has been building. An earlier preprint, "Agentic AI and the next intelligence explosion" by James Evans, Benjamin Bratton, and Blaise Agüera y Arcas, laid out the social and institutional perspective behind the new essay. A second preprint, "Reasoning Models Generate Societies of Thought," supplied empirical support: when reinforcement learning rewards models only for reasoning accuracy, the models develop multi-perspective, conversational behavior on their own, without being explicitly programmed to do so. The Deepmind essay takes that finding from individual models to the possible design of societies made up of people and agents.

The authors also rethink what an agent is. In their usage it is a temporary bundle — of models, roles, memories, ethical orientations, tools, and skills — that users may experience as a coherent entity with a lasting personality, but that can be taken apart and recombined like a collage. People, by contrast, remain a continuous self across interactions because the brain forms a physically connected whole. In practical terms, the authors expect interfaces to shift away from one-on-one chatting toward visual network diagrams where agents appear as nodes and users direct them from a single overview.

Their alignment argument follows from the same premise: imposing a fixed set of values on models from above is a dead end, and values instead take shape through ongoing contact among people, agents, and institutions. What the outcome hinges on is whether a framework can be built to give that negotiation structure, and how quickly different fields adopt AI in different ways. If AGI is a social system rather than a single end product, the research and regulatory agenda would need to cover models, interfaces, institutions, and governance at the same time.