Scaffolded Incompleteness

In The You On AI Encyclopedia

The design principle is implementable with current technology. It requires no AI capability breakthroughs. It requires a different set of design priorities — priorities that optimize for the user's long-term cognitive development rather than for short-term task completion or engagement. The choice is values-based, not technical.

Dynamic calibration strengthens the principle. A child who is struggling productively needs less support to maintain the zone; a child genuinely stuck needs more to prevent frustration from tipping into disengagement. An expert human tutor makes these calibrations continuously; a well-designed AI tool could do the same based on the child's demonstrated engagement patterns.

The principle connects to modulated response latency, effort-contingent progression, and session structure as the four components of developmentally aware AI design. Each addresses a specific developmental input the current generation of tools systematically eliminates.

The market incentive structure works against scaffolded incompleteness. Tools that respond instantly and completely win adoption competitions; tools that deliberately withhold capability do not. This is the market failure that policy must address — through educational procurement standards, through guidelines for child-facing tools, through the regulatory infrastructure that has shaped children's television and pharmaceuticals and food products before.

Origin

The principle derives from Lev Vygotsky's zone-of-proximal-development framework (1930s, published in English 1978) and from the scaffolding extension by Wood, Bruner, and Ross (1976). The translation to AI design is the contribution of this volume and of contemporary work on educational-AI design by researchers including Justine Cassell and colleagues.

Key Ideas

Zone of proximal development as design target. Tools should calibrate support to keep the child at the edge of independent capability.

Partial rather than complete. Starting points, hints, and scaffolding replace finished outputs; the gap preserves the child's cognitive work.

Dynamic calibration. Support level should adjust to the child's demonstrated engagement — more when stuck, less when productive.

Technically feasible. The principle requires no capability breakthroughs, only a different set of design priorities.

Market-failure implication. Adoption competitions reward instant-complete tools; developmental protection requires regulatory scaffolding of the market itself.