Gregory Bateson used to tell a story about a man who kicked a stone and a man who kicked a dog. The stone moved according to the physics of the kick — a unilateral transaction. The dog responded according to its own internal organization — a bilateral transaction. Mary Catherine Bateson absorbed this distinction and extended it into a general theory of creation. Everything interesting in the world was dog-like rather than stone-like. Nothing interesting was produced unilaterally. Everything interesting was collaborative, in the deep sense that its properties emerged from the interaction rather than from any single participant. This framework dissolves the anxiety about AI and creativity that pervades the cultural discourse — an anxiety that typically assumes a unilateral model of creation that never actually existed.
The anxiety about AI and authorship assumes the artist has an idea, realizes it through skill and effort, and produces a result that is the artist's creation — a product of the artist's mind, bearing the artist's signature, belonging to the artist in the way that a stone's trajectory belongs to the kicker's foot. AI threatens this model because it provides an alternative source of realization. Bateson's collaborative framework dissolves the anxiety by dissolving its premise. Creation was never unilateral. The artist never kicked a stone. The artist always kicked a dog — always engaged with a medium that responded according to its own internal organization, always participated in a bilateral exchange in which the properties of the result emerged from the interaction rather than from the artist's intention alone.
The painter's brush resists. The marble has a grain. The language has a syntax that pushes back against the writer's meaning, that bends the sentence in directions the writer did not intend, that sometimes produces felicities that the writer's conscious intention could not have generated. The medium is always a collaborator, and the result is always a joint production. What AI changes is the sophistication of the collaborator, not the collaborative nature of the process. The brush is a simple collaborator. Language is more complex. The AI is a still more complex collaborator.
The quality of the collaboration depends on the quality of the bilateral exchange — on whether both participants are genuinely contributing their internal organization to the interaction, or whether one participant is dominating and the other is merely executing. In a rich human-AI collaboration, the human brings genuine intention — real questions, real uncertainty, real investment in the outcome — and the AI contributes its full internal organization — its capacity for pattern-finding, connection-drawing, structural analysis. The result is a joint production with properties that neither participant could have generated alone. In a poor collaboration, the human provides a perfunctory prompt and accepts the AI's output without genuine evaluation. The result is a unilateral production by the AI, with the human serving as a passive receiver.
Bateson's framework suggests that the quality of the human-AI collaboration is determined less by the sophistication of the AI than by the quality of engagement the human brings to the exchange. A sophisticated AI paired with a disengaged human produces sophisticated but ungrounded output — polished prose without genuine thought, elegant connections without real understanding. A sophisticated AI paired with an engaged human — one who brings real questions, evaluates output critically, feeds genuine judgment back — produces something neither could produce alone.
The framework descends from Gregory Bateson's 1960s cybernetic writing, particularly his essays on mind and system in Steps to an Ecology of Mind. Mary Catherine Bateson extended it through her work on joint performance, her mother-infant research, and her cross-cultural studies of collaborative creative practice.
The framework received direct application to AI in Bateson's late-career writing and in this posthumous synthesis. It offers one of the few fully worked-out philosophical responses to the authorship anxiety that has dominated discussion of AI and creative work.
Creation was never unilateral. Every creative act has always been a bilateral exchange between maker and medium.
AI changes collaborator sophistication, not collaborative structure. The exchange is richer; the fundamental nature of the process is unchanged.
Quality of human engagement determines quality of output. A sophisticated AI with a disengaged human produces ungrounded fluency; the same AI with an engaged human produces genuine joint work.
The seduction of the smooth is a collaboration failure. When the human stops evaluating and starts accepting, the bilateral exchange degenerates into unilateral consumption.