Co-Evolution of Technology and Institutions
The standard model of innovation assumes a linear sequence: technology is developed, society responds, institutions adapt. Juma's research demonstrated the relationship is not linear but co-evolutionary — technologies shape institutions and institutions shape technologies simultaneously, in a process of mutual influence that produces outcomes neither could have produced alone. The distinction is not merely academic. It determines what kind of institutional action is possible and when. Under the linear model, the institutional challenge is reactive — society must catch up to the technology. Under the co-evolutionary model, the institutional challenge is creative — society's institutional choices shape what the technology becomes, and the quality of those choices determines the trajectory of the technology's development for decades after the choices are made.
In the AI Story
The printing press did not arrive in finished form and wait for institutions to respond. It evolved in dialogue with the institutions that governed its use. The university's demand for standardized textbooks shaped the economics of printing. The scientific society's demand for accurate reproduction shaped typesetting technology. The copyright regime shaped the relationship between author and publisher. Each institutional choice created demand signals, constraints, and incentives that channeled the technology's development. The press that existed in 1500 differed from the press of 1450 not merely because the technology had improved but because the institutional environment had changed, and the changed environment had shaped the technology's evolution.
The co-evolutionary perspective reveals something the linear model conceals: there is a window of maximum institutional leverage during every innovation transition, and the window corresponds to the early period when the technology has not yet stabilized, the institutional environment has not yet crystallized, and the range of possible developmental trajectories remains wide. As co-evolution unfolds, the range narrows. The technology stabilizes around specific architectures. The institutional environment develops norms calibrated to those architectures. The cost of changing direction increases. This narrowing — what scholars of technology call lock-in — means institutional choices made during the early window have disproportionate influence on long-term outcomes.
The AI transition is currently in this window. The technology has not stabilized. Multiple architectural approaches are competing. Multiple deployment models are being tested. Multiple governance frameworks are being proposed. The range of possible trajectories is wide. This means institutional choices made now — about education, regulation, professional norms, organizational design — will have outsized influence on what AI becomes over the coming decades. The choices will be encoded in the technology's architecture, in the institutional environment's norms, and in the feedback loops that connect them. Once encoded, they will be costly to reverse.
The co-evolutionary framework also challenges the convenient fiction of technological neutrality. If technology and institutions co-evolve, then the technology itself embodies institutional choices — the design decisions made during development, the values embedded in the training process, the assumptions about use that shape the interface, the economic models that determine pricing and access. The technology is not a neutral tool waiting to be governed. It is a participant in governance, carrying within it the institutional choices of its creators. AI tools developed in Silicon Valley carry within them the assumptions of the Silicon Valley institutional environment: reliable connectivity, English-language communication, formal employment relationships, legal protections for intellectual property, venture capital as primary funding, and a cultural orientation toward speed and disruption. These assumptions are not neutral technical features. They are institutional choices embedded in the technology's architecture.
Origin
The framework emerged from Juma's engagement with the technology studies literature of the 1980s and 1990s — particularly Bijker, Pinch, Callon, and Latour's social construction of technology work — integrated with his own empirical research on agricultural innovation. The co-evolutionary framing allowed him to move beyond both technological determinism and social constructivism to a position that acknowledged the agency of both technology and institutions.
Key Ideas
Mutual influence, not linear sequence. Technology and institutions shape each other simultaneously, producing outcomes neither could produce alone.
Creative rather than reactive. Institutional choices during transitions are constitutive of what the technology becomes, not merely responsive to what it already is.
Window of maximum leverage. Early-transition institutional choices have disproportionate long-term influence because they shape the co-evolutionary trajectory before lock-in occurs.
Technology is not neutral. Technologies carry within them the institutional choices of their creators; the deployment context determines whether those choices fit.
Urgency of the window. The AI transition is currently in the window of maximum leverage; choices made now will compound for decades.
Appears in the Orange Pill Cycle
Further reading
- Calestous Juma, Innovation and Its Enemies, ch. 10
- Wiebe Bijker, Thomas Hughes, and Trevor Pinch, eds., The Social Construction of Technological Systems (MIT Press, 1987)
- Bruno Latour, Reassembling the Social (Oxford University Press, 2005)
- Andrew Feenberg, Between Reason and Experience (MIT Press, 2010)