The job being taught

In South Philadelphia, about 20 adults sit in librarian Charlie Bailey's hour-long workshop. Bailey first explains how AI chat tools work and why someone might want to use them or abstain; then he walks through the common platforms and devices one by one, projecting the steps for turning features off. The workshop is called Avoiding AI.[1]

Seen from the street, the job description here is unusual: the product being taught is the settings screen. The clearest evidence is the kind of question Hannah Cyrus, who works at Bangor Public Library in Maine, keeps getting. People increasingly ask her how to turn this stuff off, why it is trying to write their emails for them, why it is summarizing a one-sentence email. When switching a feature off takes more work than learning the product, the design has already made the user's decision.[1]

The demand is real, but the numbers come from two libraries

The comparison has to be made against each library's own baseline. Cyrus's Intro to Computers classes usually draw about a dozen people; interest in Avoiding AI was heavy enough that she capped in-person registration at 30, opened a waitlist and streamed the session on Zoom. Including the livestream, about 70 people attended each of the first two. After she wrote about the format in a professional journal, dozens of librarians around the world contacted her; Bailey is one of them. His library's Instagram post drew more than 2,000 likes and 220 shares, where its posts mostly stop at a few dozen likes, and a second session was added because of the demand.[1]

These numbers work as a measure of difficulty, not of prevalence. Two libraries, a few hundred people and one social post cannot show how many people actually switched anything off. The same attention has another possible explanation: a librarian criticizing Big Tech is itself attention-worthy, and turnout at a free local class shows that a topic is current, not that phone habits have durably changed.[1]

Switching off is not refusing

The mood in the room says the same. Cyrus points out how useful optical character recognition is for scanning old documents at the library; one attendee says she is not against AI when it comes to medical advances. The tip shared in the room is about changing a setting rather than leaving the technology: a parameter appended to a Google search hides AI results. The concrete signal to wait for is whatever would make the workshop unnecessary: one of the large platforms publishing, by the end of the year, a single easily found control that turns device-level AI features off in one place. When that setting arrives, this class will be about something else.[1]