School EdTech now matters more than the school logo

Zvi Mowshowitz’s Childhood and Education #20 looks at school screens, EdTech and phone bans. The useful question is not whether children should ever use screens. It is who controls learning when software starts running the school day.

Choosing a school increasingly means choosing its software

Mowshowitz starts from a practical observation: when parents pick a school, they often pick an educational software stack too. The child is not only working with a teacher. They are working inside a system that sets pace, repetition, rewards and punishments.

His main example is Ryan Moulton’s account of i-Ready. Moulton describes a child who liked math, then began dreading it after being pushed through the school’s software. He says other parents reported similar reactions: fatigue, tears and children trying to avoid the system.

AI tutors inherit the trust debt of today’s platforms

That is uncomfortable context for AI companies. A personalized tutor sounds better than one lecture for a whole classroom, but schools have already bought tools that promised individualization and behaved like cheap control layers.

The difference between a useful AI tutor and bad EdTech will not be that one uses a model. It will be workflow design: whether the child understands the task, whether an adult can inspect what happened and whether the system allows a change of route when it is clearly doing harm.

Automation can enforce a bad rule without getting tired

The central risk is not the screen itself. It is software that cannot tell when a rule has become stupid. A teacher usually would not force the same instruction dozens of times in a row, because resistance, boredom and lost meaning are visible. Software can do it precisely and without embarrassment.

That matters for AI in schools too. A model may explain better than an old drill product, but if it becomes a mandatory mode with no exit, it repeats the same failure behind a smarter face.

Adult override will matter more than the demo

For school AI tools, the signals worth watching are less flashy than demos. Teacher audit trails, the ability to skip a bad task, parent visibility and clear data rules will say more than a polished ideal lesson.

The next question is who pays for bad deployment. If broken software makes the child look like the problem, the school has a governance failure. If the system can be stopped, explained and replaced by human judgment, the AI tutor has a chance.

Lilith's verdict

EdTech has a trust problem: the parent at the kitchen table is not studying the roadmap, but a child who does not want to hear the word math after school. The AI tutor earns its place only when an adult can close the notebook in its face.