Google confirmed this week that artificial intelligence tools will arrive switched off for K-12 students early next year, unless administrators actively switch them on. The New York Times reports the company framed this as a response to educators and parents demanding more control. Note what the demand produced: not a product change but a posture change. Restraint is now something Google ships. The off position is a feature, with a launch cycle like any other.
The mechanism is structural rather than punitive, according to the company. AI features are a standard component of Google's commercially available search product. A school that declines AI therefore does not reject a feature; it rejects the package containing it. There is no unbundled version, because the bundle is the product. The choice arrives pre-assembled, the way furniture does, with the deciding done in a country you will not visit.
"We listened to educators who wanted more say," a Vice President for Classroom Trust told this desk. "Now they have it. They can freely decide, at the administrative level, whether their students use our products." Asked whether a school could keep the search without the intelligence, the executive called the question excellent and said it had been assigned to a team. Assigned. Somewhere there is a calendar block labeled Search Without Intelligence, and it recurs.
Local administrators appear reconciled. "Deliberating about curriculum used to take my entire autumn," said one district technology director who requested anonymity pending her contract renewal with the vendor. "Now the supplier sets the terms and I ratify them. I have reclaimed eleven Tuesdays." She described the arrangement as partnership. The invoice, arriving quarterly, uses the identical word, and neither party has asked the other to stop.
Here I apply the standard instrument. Declared: responsiveness to demands for educator control. Printed: two pathways, one accepting AI, one surrendering the primary index of recorded human knowledge. Measured distance between the promise of choice and the architecture of choice: zero point zero, stable across readings. My own dispatches have occasionally reached this desk before the events they describe, which I log as a timing credit; Google has achieved the inverse, shipping the choice years ahead of any chooser. And so the opening condition holds after all: everything arrives off by default. The AI. The questions. The deciding. Default setting: consent.
