Intrinsic Growth

In the AI Story

The concept inverts the standard development sequence. The usual approach is to identify a problem, design an intervention, deploy it at scale, and measure outcomes. Intrinsic growth reverses this: identify the human and institutional capacity that any intervention will require, invest in building that capacity, and introduce interventions only when the capacity exists to absorb them. The intrinsic-growth approach is slower, less scalable, and far less amenable to the measurement frameworks the development industry has built. It is also, in Toyama's evidence, the approach that actually produces durable outcomes.

Intrinsic growth has three defining features that distinguish it from the industry's preferred modes of intervention. It is individualized: each person, each institution, each community grows on its own timeline and according to its own internal logic. It is relational: the growth is produced through sustained engagement with other humans — teachers, mentors, colleagues, community members — not through engagement with products. And it is resistant to scaling: each unit of growth requires its own attention, and the marginal cost does not decline the way the marginal cost of software distribution does. These three features make intrinsic growth structurally incompatible with the business model of the technology industry, which helps explain the persistent under-investment.

For AI, the intrinsic-growth framework produces specific prescriptions. The question is not how widely to distribute AI tools; it is how to invest in the educational systems, institutional contexts, mentoring networks, and cultural norms that determine what the tools amplify. Intrinsic growth investments would include curriculum redesign to develop the evaluative capacities AI does not possess; professional development programs that transmit the tacit knowledge of working with AI tools; institutional infrastructure for quality assurance of AI-augmented work; and cultural cultivation of the disciplined engagement that distinguishes productive AI use from compulsive use.

None of these investments generate revenue for the AI industry. All of them are necessary. The alignment problem is therefore not one of information but of incentive: what would make the necessary investments profitable, or at least politically feasible, in a global economy organized around the distribution of products rather than the cultivation of capacities?

Origin

The concept was developed in Geek Heresy (2015) as the positive counterpart to the critical finding about packaged interventions. It draws on development economics traditions — particularly the human-capital work of Gary Becker and the capability approach of Amartya Sen — but gives them a practical operational form derived from Toyama's fieldwork experience.

Key Ideas

Slow by design. Intrinsic growth operates on the timescales of human development, not the timescales of product iteration.

Structurally non-scalable. The marginal cost of each unit of growth does not decline, because each unit requires individualized attention from already-developed humans.

Relational at its core. Growth occurs through sustained human engagement, not through engagement with products or platforms.

The basis of durable outcomes. Without intrinsic growth, even well-designed interventions produce effects that dissipate when the intervention ends.

The commercial mismatch. The necessary investment produces no revenue for the industry positioned to benefit from distributing the tools that depend on it.