You On AI Field Guide · The Open Future The You On AI Field Guide Home
TxtLowMedHigh
CONCEPT

The Open Future

Smolin's physical claim that the future is not contained in the present—that time is real, the next state of the universe is not determined by its current state, and the dams builders construct in the river of intelligence are therefore constitutive rather than merely corrective.
The dominant framework for thinking about AI development is implicitly Newtonian: the architecture is fixed, the training data determines the outputs, capability scales deterministically with resources, and the future of AI is already implicit in the scaling curves of the present. More data and more compute yield more capability; the trajectory is determined by the initial conditions. Lee Smolin's physics denies this comprehensively. If time is real—if the future is not contained in the present but is genuinely, ontologically open—then the scaling hypothesis is not a law but an observation about what has happened so far, extrapolated forward on the assumption that the future will resemble the past. Phase transitions violate this assumption by definition: the system reorganizes, and the rules that governed the previous phase no longer apply. The open future is the physical ground for the cycle's most demanding ethical claim: if the future were determined, the dams that [YOU] on AI advocates would be adjustments to a predetermined outcome. If the future is genuinely open, the dams are constitutive—they do not modulate a trajectory that exists; they create a trajectory that would not otherwise exist. The builder's responsibility is not limited to optimizing a predetermined outcome. It is total: what gets built is what the universe becomes.

Origin

The open future is the ethical corollary of Smolin's thesis that time is real and fundamental rather than an emergent illusion derived from deeper, timeless physics. If the laws of physics are genuinely time-reversible at the fundamental level, then past and future are ontologically equivalent—the block universe of Einsteinian physics, in which every moment exists with equal reality and the apparent flow of time is a feature of perception rather than of reality. In this framework, the future is already 'there,' implicit in the present configuration, waiting to be discovered rather than created.

Smolin argues that this picture is wrong not just empirically but conceptually: a timeless universe has no explanation for why it has the laws it has rather than any other laws. If the laws evolved through temporal processes—through cosmological natural selection—then both the laws and the universe's specific trajectory are genuinely temporal, genuinely contingent, and genuinely open. The next state of the universe is not written anywhere. It comes into being through the temporal process itself, in the thick present where the past has been determined and the future has not.

Key Ideas

Constitutive versus corrective action. The difference between a determined future and an open future is the difference between corrective and constitutive action. If the future is determined, the builder's task is to accelerate the predetermined outcome or, at best, to modulate its speed. If the future is genuinely open, every choice is constitutive: it does not adjust a trajectory that exists; it creates a trajectory that would not otherwise exist. The dams that protect the development of human capability, the architectural choices that preserve genuine cognitive friction, the professional norms that maintain the value of deep engagement—none of these are adjustments. They are acts of creation in a universe that takes time seriously.

The paradigm shift the AI industry needs. The Newtonian assumption inside AI development—that the scaling curves determine the trajectory, that more compute necessarily produces more capability, that the future is implicit in the present configuration—is precisely the fishbowl that Smolin spent his career cracking in physics. Inside the fishbowl, responsibility is limited: if the future is determined, the builder's task is to execute the inevitable more efficiently. Outside the fishbowl, responsibility is total: the choices made during the current phase transition shape what the universe becomes, not what version of a predetermined future arrives sooner.

Recombination versus genuine novelty. The open future is the physical ground for the distinction between recombination—finding new arrangements within an existing space of possibilities—and genuine novelty—expanding the space of possibilities itself. Current AI systems operate deterministically on fixed datasets and therefore explore existing possibility spaces without participating in the thick present where genuine novelty enters the universe. The human contribution of biographical specificity, genuine questions, and the open future is not a residue or a luxury. It is the epistemologically necessary contribution of the element in the collaboration that has access to the thickness of the present.

The baby meeting the novel. Smolin offered a striking analogy at a 2023 symposium: babies do not attempt to predict who they will encounter next. They engage sequentially, asking 'Who is that?' after each new meeting. Each encounter is genuinely new. A prediction machine operates on the Newtonian assumption: the future is implicit in the data, and the machine's job is to extract it. A construction machine operates on the temporal assumption: the future does not yet exist, and the machine's job is to participate in its creation. The distinction Smolin draws between these two roles is the deepest description available of what distinguishes the current AI paradigm from what a genuinely temporal AI might become.

Explore more
Browse the full You On AI Field Guide — over 8,500 entries
← Home0%
CONCEPTBook →