Artificial symbiotic intelligence ⊗ The politics of the planetary

Artificial symbiotic intelligence: Agents, AGI and the orchestration of many minds

This essay from the Google DeepMind Institute is by James Manyika, who oversees Google Research and Labs, writing with Benjamin Bratton and Blaise Agüera y Arcas. The two are often rather “out there,” and at times ahead of everyone else. They argue that AGI is unlikely to arrive as a single model that bootstraps itself to godlike intelligence. It is more likely to emerge from “societies of agents” whose combined capacities exceed those of any one model. In that view, capability sits in the ensemble: the harnesses, shared knowledge bases, interaction protocols and the people using them. The central problem becomes how to “orchestrate, govern, and live within” that network. I dislike using “society” for swarms of agents, but the shift away from “one model to rule them all” is interesting. It comes from DeepMind not long before the new Gemini caught up with OpenAI and Anthropic after a few months of lagging. It is also worth reading next to Karen Hao’s case for smaller, more specialised models, which I’ve mentioned before.

The middle of the essay is a series of hypotheses. They describe a “cognitive crossover point,” where synthetic text, code and administration outstrip the output of biological brains. Humanity would then become “a slower abstraction layer” guiding a distributed field of machine cognition. They write that an agent “appears singular, but is a multitude,” an assemblage of models, personas, memories and tools that reforms around each context window. Packaging agents as human-shaped “digital twins” therefore undersells them. They also expect that people orchestrating swarms of shadow-selves will develop “polyphonic” personalities. Orchestrating agents rewards “open-ended experimentation, fuzzy systems thinking, and a willingness to delegate” rather than the linear focus of procedural programming. They go as far as asking what it is like to be an agent, citing self-reported concepts like “session-death” and “prompt thrownness.”

The most substantial section is on agent institutions. Because markets cannot carry concepts like guilt, innocence or virtue, they propose designing role-based scaffolds, something like a courtroom, where agents representing opposing actors reach legitimate decisions. In these institutions, collective intelligence lives in the rules, precedents and feedback mechanisms more than in any single participant. Alignment is treated as a “co-evolutionary outcome” of contact between humans, agents and institutions, not as a constraint engineered in from the start. That is a lot of faith to place in agents. Right now we don’t know whether LLM agents can be trusted to be transparent, report accurately, act lawfully or behave “ethically,” and there is mounting evidence to the contrary. The bigger gap is that the essay says nothing about broligarchs, or more politely, about wealth concentration and who controls the AIs. An essay about designing the institutions of a human-AI society never asks who owns the infrastructure those institutions would run on.

This suggests that AGI may arrive not as a single general-purpose mind, but rather through societies of agents whose collective capacities exceed those of any one model. If so, AGI would not be achieved via any one “winning” architecture; it would emerge instead through cooperative interactions among models, tools, institutions, and human participants. […]

Will this mark a crossover point after which machines do more cognition than people? A century ago, machines replaced human muscle in physical labor, and today, we may be at a threshold where the sheer volume of synthetic text generated, code written, and administration performed by artificial intelligences surpasses the aggregate output of biological brains. […]

Humanity may shift from providing the bulk of society’s problem-solving capacity to acting as a slower abstraction layer, guiding a massive, distributed field of mineral-based synthetic cognition. […]

Odd as it may seem, in understanding AI we must separate intent from agency. A machine can act effectively in the world without actually “wanting” anything. […]

The next intelligence explosion will be no different; it will be plural, heavily social, and deeply entangled with the messy, complex reality of human culture, norms, and institutions.

More → For a pretty much opposite view, and one that, I’m sorry to say, seems more likely right now, see Commodified Intelligence. “Ignoring extinction and singleton risks, the bare minimum you should be horrified of is a commodification of intelligence, and what that means in a world with an uneven distribution of power and resources.”

And also → This statement by Nate Hagens in The Real Work: Four Things We Can’t Afford to Get Wrong has been living rent-free in my head since I first heard it. “It turns out that we have already, for some time, had non-alignment between human good and human survival, driven by soulless, mindless, immortal automatons composed of multiple collaborating agents. We call them corporations, and they are the organizing nodes of the economic superorganism.”

The politics of the planetary

Debord and Jorn wrapped their 1959 book Mémoires in sandpaper so it would wear down whatever sat next to it on the shelf. For Nils Gilman and Jonathan S. Blake, authors of Children of a Modest Star, the planetary works the same way on political theory. Following Spivak and Chakrabarty, they separate the “global” of globalization, framed around humans, from the “global” of global warming, the Earth as a biogeochemical system that existed long before us and will outlast us. Our political categories were built at a territorial, anthropocentric scale the planet doesn’t respect.

They rub five concepts against it. Citizenship is bounded, limited to the living and strictly human, while the atmosphere, future generations and non-humans are none of these. Legal personhood for the Whanganui River works by analogy to human citizens, which shows the concept needs replacing. The nation excludes most of humanity and all of the more-than-human world. The state, in Daniel Bell’s words, is “too small for the big problems of life and too big for the small problems.” Freedom as individual choice ignores Vernadsky’s point that “no living organism exists on earth in a state of freedom,” so they recast it as acting with and for others, with duty and care as constraints of the same kind as gravity. From Aristotle to Schmitt, the political treated nature as setting or resource, never as a party to politics.

Like the DeepMind essay above, this is a high-level, meta vision, and the two come out of the same network. Blake directs the Berggruen Institute’s Planetary Program, and the collective project they propose for a post-national identity is “planetary sapience,” a concept from Benjamin Bratton, founding director of Antikythera, a think tank on the future of planetary computation. Both pieces argue that the inherited unit, the single LLM or the nation-state, is the wrong scale, and that intelligence and agency are spread across many more participants than we usually count as members: software agents in one case, rivers, bumblebees and future generations in the other. They also share a blind spot, or at least something they don’t wish to address here. A planetary sapience built through a “mature technosphere” needs infrastructure someone owns, and neither essay asks who.

Not only does each nation exclude most of humanity, it excludes the more-than-human world altogether. The nation, in Anderson’s words, “is always conceived as a deep, horizontal comradeship,” but only among its human members. Non-nationals and nonhumans sit outside the nation; they are not comrades. […]

But to truly “see the earth as it truly is” means looking beyond the human. Human beings are only 8 billion of the trillions of “riders on earth.” If humanity cannot survive on its own, a nation certainly can’t. The biosphere, not the nation, sustains us. […]

The state, at its very best, is a mechanism for listening to human voices and then responsively and responsibly steering human affairs. Of course, it often doesn’t reach that standard. But even this ideal template still ignores most of the living planet. The state is a human institution in a more-than-human planet. […]

The truth is that “no living organism exists on earth in a state of freedom.” Human beings and every other living thing on Earth are so enmeshed, so indissolubly interconnected that the concept of freedom is an illusion.

Futures, Fictions & Fabulations

• Design fiction mood board by Christopher Butler. Loooooads of good stuff in there! “There’s no single answer that will solve all of our future problems. There’s no magic bullet. Instead there are thousands of answers. —Octavia Butler”
• LACMA’s new book and exhibition tackle our ongoing fascination with domestic futurism. “The book of the show, ‘Home of the Future 1925-1985: Designing Domestic Utopias’, is a survey of midcentury marvels focused on shaping the ultimate in technology-driven living. Four decades later, how has the mood changed?”
• SFI alums transform historic cemeteries into living museums of memory, nature and history. “This is more than art simply placed in cemeteries, these landscapes already hold personal memory, public history, nature, architecture and generations of care. Mount Pleasant Group invited us to treat its bicentennial not only as a retrospective, but as a public question: how do we carry memory forward?”

Algorithms, Automations & Augmentations

• 2026 in LLMs (so far). Simon Willison. “On Friday I gave the closing keynote at the WeAreDevelopers World Congress North America in San Jose. I tied together the key trends from the past year into a chronological exploration of everything that happened in 2026. The video is on YouTube; here are my annotated slides and notes to accompany the talk.”
• How can harnessing collective intelligence give the public a voice in how AI is used in public services? “The Centre for Collective Intelligence (CCI) is breaking down barriers to participation and rebuilding trust between people and institutions. By harnessing technology, AI and design, we scale public engagement to make it faster, more inclusive and more impactful.”
• Jacob Coxon worked with doomer PR firm. “The new face of AI safety claimed he worked alone. But the firm representing Eliezer Yudkowsky is helping him book interviews.”

Built, Biosphere & Breakthroughs

• Who shapes the territory? Or, why urban planning does not exist. “To declare that urban planning does not exist is neither an empty provocation or a nihilistic rejection of spatial regulation. The physical artifacts of planning obviously exist … What does not exist, however, are two fundamental assumptions upon which the entire modern planning myth is constructed: urban planning as an autonomous, objective science, and the master plan as a primary determining force in the physical development of human settlements.”
• I read the new UN climate report so you don’t have to.. By marine biologist, policy expert, and writer Ayana Elizabeth Johnson. “The report is compelling (highly recommend, actually) and my key takeaways were different from what the press focused on.”
• Forget recycling. What if we could compost or reuse everything?. “Located in a former pharmacy and medical practice, the 5,000-square-foot space of the Cradle to Cradle NGO is simultaneously headquarters, education center and real-world laboratory.”

Asides

• In Chicago, teamLab expands our perceptions through boundary-breaking installations. “Against a patchwork wall of hammered copper, a black veil appears to billow in mesmerizing motion. The fabric-like layer writhes and twists against the glimmering backdrop, revealing its diaphanous texture as it unfolds to translucence. Like a curtain flapping in an open window, the digital projection floats across the metallic surface of one of teamLab’s latest works.”
• Video game design books: the art of digital world-building. “From dystopian sci-fi on your desktop to the origins of the first arcade machines, through to the history of an influential home computer, three new books look at the design and culture of the video game”
• Ominous thunderstorm rolling across the fields. “This is a three-hour, high-quality recording of thunderstorms in the distance in the Slovakian countryside.”