Pro-Innovation Bias
Pro-innovation bias is the assumption that an innovation should be diffused to and adopted by all members of a social system, that it should be diffused rapidly, and that it should be neither reinvented nor rejected. Rogers identified this bias as the most persistent and least recognized distortion in diffusion research — and in his own earlier work. The bias operates not through deliberate advocacy but through structural position: the people who study and promote innovations are, by definition, people who have engaged with those innovations deeply enough to find them interesting, and that engagement systematically disposes them toward favorable evaluation. Rogers's career-long push against this bias is one of his most important legacies, and the one most urgently needed in the AI transition.
In the AI Story
The bias manifests in several ways. It privileges studies of successful diffusions over studies of failed or harmful ones. It treats non-adopters as problems to be solved rather than as rational actors making rational decisions. It underweights consequences — particularly undesirable and indirect ones — in favor of adoption metrics. It generalizes from the experience of enthusiastic adopters to populations whose circumstances differ.
Rogers recognized that his own early work exhibited the bias. The fifth edition of Diffusion of Innovations includes explicit self-criticism and a sustained argument that researchers must attend to the structural conditions that produce apparent success for some populations while generating documented harm for others.
The AI transition is unusually susceptible to this bias because of a structural feature Rogers did not encounter in his empirical work: the innovation participates in its own advocacy. The most compelling accounts of AI's transformative potential are produced by people who are themselves transformed — whose experience of the orange pill moment makes their advocacy more vivid and, simultaneously, more structurally biased.
Edo Segal acknowledges this explicitly in The Orange Pill: builders have a blind spot; they see what the tool makes possible and structurally undercount the cost. Rogers's framework provides both the vocabulary for this acknowledgment and the corrective — the insistence that laggards' resistance contains information that innovators cannot generate themselves.
Origin
Rogers introduced the concept of pro-innovation bias in the 1983 third edition of Diffusion of Innovations, acknowledging that his own earlier work had exhibited the bias and that the field as a whole needed to reckon with it.
The self-critique deepened in subsequent editions, drawing on international development research that documented how celebrated agricultural modernizations had produced devastating outcomes for the populations they were supposed to help.
Key Ideas
Structural, not intentional. The bias operates through the researcher's position, not through deliberate advocacy.
Makes non-adoption invisible as data. Laggards' resistance is treated as a problem to solve rather than as information about costs the innovators cannot see.
Particularly intense in AI. The innovation participates in its own advocacy, intensifying the structural bias.
Self-correction is essential. Rogers insisted that researchers must actively counter the bias through attention to consequences and to the structural position of late adopters.
Appears in the Orange Pill Cycle
Further reading
- Rogers, Diffusion of Innovations (2003), Chapter 3
- James Scott, Seeing Like a State (Yale, 1998)
- Edward Tenner, Why Things Bite Back (Knopf, 1996)