George Basalla

In the [YOU] on AI Field Guide

The cycle identifies December 2025 as a threshold—the moment Claude Code crossed a capability boundary that made the previous paradigm categorically different. Basalla's framework does not dispute that the capability gains were real. It disputes the framing. What appeared to the user as a phase transition was, from the perspective of the engineering lineage, the most recent increment in a continuous process of accumulation. The user experienced the threshold because the user encountered the artifact at the moment when the accumulated increments crossed a perceptual boundary. But the boundary was in the perception, not in the process. The process was continuous throughout. And understanding it as continuous is essential for responding to it wisely.

Basalla's most practically useful insight for the cycle is about the selection environment. The discourse around AI focuses almost exclusively on capability benchmarks—which model scores highest, which generates the fewest hallucinations. Basalla's framework says this is the wrong place to look. The decisive factor is not which AI system performs best in the lab but which AI system fits the selection environment in which it must survive—the institutional inertia of large organizations, the cultural narratives about what AI should be, the regulatory frameworks being written now, the economic path dependence created by early-mover advantages. A technically superior AI system that requires organizational restructuring, professional retraining, and regulatory navigation faces a hostile selection environment regardless of its benchmark performance.

The cycle's interest in fantasy inventions—the stories cultures tell about technology before the technology arrives—maps directly onto Basalla's analysis. The autonomous-mind fantasy that saturates popular representations of AI, from HAL 9000 to Samantha in Her, has created a cultural niche that AI systems must fit to gain adoption advantages that have nothing to do with their technical properties. The fantasy precedes the artifact. The artifact fills the niche the fantasy created. This is why conversational AI reached fifty million users in two months—it entered a selection environment primed by decades of imagining exactly this kind of system. And it is why the most beneficial AI architectures—those oriented toward collaborative partnership rather than autonomous agency—may face a selection environment less receptive than their social value would warrant.

Basalla's anti-heroic perspective on the inventor's illusion is the cycle's sharpest corrective to the mythology surrounding AI's emergence. The concentration of narrative in a small number of founder-figures produces a concentration of perceived agency that disempowers the people who actually shape the selection environment: regulators, educators, managers, parents, workers. Basalla's framework distributes that agency back outward, to the institutional forces that determine which variations survive. The environment is the decisive variable. The environment is made by everyone. And the recognition that this is so is the first step toward building the structures that direct the transition toward human flourishing.

Origin

Basalla was born in 1928 and trained at Harvard under I. Bernard Cohen, the pioneer of the history of science in the United States. His 1967 paper 'The Spread of Western Science,' published in Science, displayed the characteristic features of the evolutionary framework two decades before its formal statement: attention to the mechanisms of transmission, resistance to heroic narratives, interest in the social and institutional conditions that determine which scientific practices take root in which environments.

The Evolution of Technology, published in 1988 by Cambridge University Press, was the single volume that consolidated the framework. It was not a popular book in the conventional sense; it became something more durable: a framework. The four pillars he identified—diversity, continuity, novelty, and selection—organized a sustained, evidence-based assault on two hundred years of heroic-inventor mythology. The assault was methodical and grounded: simultaneous invention as structural feature of the process, not cosmic coincidence; artifact lineage as the primary unit of analysis; the selection environment as the decisive variable.

Basalla acknowledged without fully exploring a recursive dynamic that AI makes unavoidable: the artifact, once created, alters the environment that selects future artifacts, including the cognitive architecture of the maker. The smartphone restructured attention. AI restructures thinking—the process by which human beings form judgments, generate ideas, and make decisions. The recursive loop now operates at the level of the cognitive architecture itself, and the depth of the reshaping is what makes the AI transition categorically more complex than any prior transition his framework was designed to analyze. He died in September 2025, three months before the threshold event that would have given his framework its most consequential test.

Key Ideas

The Continuity Thesis. Every artifact descends from a prior artifact through an unbroken chain of modification. The continuity thesis is Basalla's most defiant claim, because it directly contradicts the narrative that most people carry about how technology works. The transformer architecture was a variation on existing attention mechanisms. RLHF was a variation on existing reinforcement learning. The natural language interface descended from decades of NLP research. At no point in the lineage of any AI system does a discontinuity appear—what appeared to users as a phase transition was the cumulative effect of incremental improvements crossing a perceptual threshold.

The Selection Environment. The survival of a technology has almost nothing to do with technical superiority. The selection environment—the economic, cultural, institutional, and regulatory forces that determine which variations persist—is the decisive variable. The electric car lost. QWERTY survived. VHS won. In each case the technically relevant factors were secondary to the institutional ones. AI's fate will be determined by the regulatory frameworks being written now, the cultural narratives shaping adoption expectations, the economic path dependence created by early movers, and the institutional inertia of the organizations that must adopt the technology.

The Inventor's Illusion. The popular narrative attributes technological change to identifiable individual geniuses—the heroic inventor who creates something from nothing. Basalla demonstrated that this narrative is a retrospective construction that serves legal, corporate, and national mythology rather than historical accuracy. Simultaneous invention—Bell and Gray filing telephone patents on the same day; Darwin and Wallace arriving at natural selection independently—is a structural feature of the process, not a coincidence. It occurs because the variation landscape constrains possible novelty tightly enough that multiple independent explorers converge on the same territory. The inventor's illusion concentrates perceived agency in a few founder-figures; the evolutionary view distributes it back to the institutional environment.

Fantasy Inventions. Artifacts imagined in fiction, mythology, or speculative thought long before the technical means to build them existed shape the selection environment by creating cultural niches that real artifacts fill. Fantasy inventions do not predict the future; they shape it, by directing inventive effort and priming the adoption environment. The autonomous-mind fantasy that runs from Frankenstein through HAL 9000 to contemporary AI discourse created the cultural reception context for conversational AI. It also created expectations that may misalign with the AI architectures most beneficial for human flourishing.

The Artifact Evolves Its Maker. Basalla acknowledged a recursive dynamic his framework did not fully explore: the artifact alters the selection environment, which includes the cognitive and social architecture of the maker. The power loom reshaped manual labor; the telephone reshaped communication; the smartphone reshaped attention. AI reshapes thinking itself. The recursive loop now operates at the level of the cognitive architecture, meaning that the selection environment for future AI will be shaped by minds that AI has already modified. The institutional structures that mediate this recursion—what Basalla called the institutional ecology of artifacts—determine whether the loop produces more capable makers or more dependent ones.