Petlibro's new Granary 2 feeder uses a built-in scale and, on the premium model, an AI camera to track exactly how much your cat eats, when it eats, and which cat is eating it. The pitch is health and care. The mechanism is continuous biometric surveillance of a being that cannot consent and would not understand the question. This week The New Yorker ran a piece asking whether the classroom should be more like the gym, meaning: should education stop being about measurable outputs and return to being about process for its own sake. These two stories do not seem like they belong in the same sentence. They are the same argument from opposite directions.
Measurement as Proxy for Love
The cat feeder exists because cat owners want to express care and the most legible modern form of care is data. If you know your cat's intake down to the gram, you are a good owner. The measure becomes the relationship. Dan Rockmore's essay in The New Yorker makes the symmetric point about AI and education: as AI presses harder on the academic enterprise, the classroom risks becoming a place where measurable performance is the only thing that can be defended, because it is the only thing the new infrastructure can see. A 2023 paper in Learning and Instruction by Drachsler and Schneider found that learning analytics dashboards in educational settings systematically increase student anxiety about performance while producing marginal improvements in actual learning outcomes. The cat does not know it is being monitored. The student does. That asymmetry may not matter as much as we assume.
The Gym Logic and the Feeder Logic
Rockmore's gym analogy is that a gym works because the process has intrinsic value: you go, you lift, you leave, the benefit accrues whether or not someone grades it. The classroom used to operate this way. So did feeding a pet: you filled the bowl, the cat ate, the relationship was in the ritual. The AI feeder replaces the ritual with a dashboard. The AI classroom replaces learning with a performance. Both upgrades are sold as improvements. Both extract something that did not have a name until it was gone. Eugenia Kuyda's conversation at Culture Slop on shaping software rather than being shaped by it is the useful counter-position here: the question is not whether to use the tool, but whether you remain the one deciding what the tool is for. The cat cannot ask that question. The student, and the teacher, still can.