The deliverer model's internal logic was coherent: content knowledge plus pedagogical technique produced adequate instruction, which could be measured through standardized assessment, which justified the system. Teacher training programs were designed around this logic, selecting candidates for their convergent academic performance and training them in delivery methods. The credential that certified a teacher certified her capacity to deliver, and the career track that followed rewarded the same capacity through increasing seniority and responsibility.
The mentor model has always been present in education, but as the exception rather than the norm. Robinson's three decades of fieldwork documented mentors operating within deliverer systems, often despite institutional resistance. The extraordinary teachers he observed shared a common practice—the capacity to see the student, to recognize what the standardized curriculum did not measure, to create the conditions under which the child's particular form of intelligence could emerge. The practice could not be trained into a teacher who had not first been mentored herself.
AI's natural language interface has destroyed the economic rationale for the deliverer model. The machine delivers content more efficiently than any human teacher—available at all hours, infinitely patient, adapted to the student's pace, responsive to her questions. If the teacher's primary function is content delivery, the teacher has been replaced. If the teacher's primary function is mentorship—the specifically human relationship through which a young person's capacities are developed—the teacher has been liberated from the mechanical labor of delivery and freed to do what only a human being can do.
The seeing of the student that mentorship requires cannot be performed by a machine. Not because the technology is insufficiently advanced but because the operation requires being a human being—carrying the accumulated understanding of what it means to be a child trying to figure out who you are, noticing the moment when a student's engagement shifts from dutiful to genuine, knowing when to push and when to wait. The decisive interventions are small and unglamorous and structurally invisible to any metric the industrial system could measure.
Robinson's distinction emerged from thirty years of observing schools that had departed from the industrial norm and produced extraordinary results. Creative Schools (2015) documented dozens of cases, with the mentor-deliverer contrast serving as the analytical backbone. The distinction drew on earlier work by John Dewey, Paulo Freire, and Maria Montessori, but Robinson gave it specific operational content derived from contemporary educational practice.
The concept's application to AI emerged after Robinson's death, as the deliverer model's obsolescence became technologically undeniable. The English teacher who stopped grading essays and started grading questions—described in the target book as a representative case—exemplifies the practical transformation that Robinson's framework now demands.
Two models, different purposes. The deliverer transmits content; the mentor develops a person. Both can produce students who perform adequately on tests, but only the second produces adults who know who they are.
AI has automated the deliverer's function. Not incidentally, not incompletely, but structurally: content delivery is now a machine capability, and the human who competes with the machine on that dimension will lose.
Seeing the student is the irreducible core. The developmental relationship requires recognition of the particular child—her form of intelligence, her patterns of engagement, her moments of breakthrough—that no measurement system can capture and no tool can replace.
Teacher training must be rebuilt. The training infrastructure that produced deliverers must be redesigned to produce mentors, requiring different selection criteria, different training methods, and different assessment frameworks than the current system employs.