Educational Kaizen for DeafBlind Learners: Artificial Intelligence as Coparticipant, Not Describer

Two literatures have grown for a decade without meeting. Educational kaizen — the transfer of the Japanese continuous-improvement philosophy into the management of teaching and learning — has produced work on course evaluation cycles, institutional quality culture, and curriculum redesign, but has addressed itself almost exclusively to sensorially typical learners. Assistive artificial intelligence for DeafBlind people has produced wearable haptic interfaces, vibrotactile finger-Braille gloves, and speech-to-Braille converters, but has framed the problem as one of transport: how to move a signal into the tactile channel. Neither literature asks what should be transported when the channel carries on the order of tens of characters per minute. This paper argues that the intersection is not a marginal application of an existing method but a case that reverses the direction of kaizen itself. Under severe bandwidth constraint the learner, not the institution, must own the improvement cycle, because no external party can evaluate the semantic yield of a tactile output. The position is derived from the protactile community’s own critique of the describer model, and the canonical tasks of educational kaizen are restated under narrow-band conditions, three of which are shown to invert.