I Was Right About AI, and Wrong About Our Children

During the 2016-2017 school year, when I was Head of School at Hillel Day School of Metropolitan Detroit, I wrote a blog called “A.I. is Coming and Why You Should Care.” I argued that artificial intelligence was reaching its tipping point and would do to white collar work what robots had already done to the factory floor. A year later, in “Utopia or Dystopia?” I predicted that a machine with human emotional intelligence would remain in the realm of science fiction. Almost ten years later and now as Head of School at Schechter Bergen, I’m looking back at what I got right and what I got wrong, and at what the research now tells us about our children.

I wrote that our children would encounter AI when they entered the workforce, and for the sixth, seventh and eighth graders whose parents were reading that blog, I was right. They’re in their twenties now, starting careers just as AI is reshaping writing, analysis and advising that are central to so many professional jobs, and the skills that I’ve argued for, for more than twenty years now, critical thinking, adaptability and creativity, matter more than ever.

But for today’s youngest children, I was wrong about the timing. AI is not waiting for them after college, because it is already here in a very big way. A RAND survey published this past March found that young people from middle school through college are increasingly using AI for homework as usage rose from 48 percent to 62 percent in just seven months of 2025.

My prediction about emotional intelligence, while technically correct as of now, was still off for all practical purposes. Machines still don’t feel anything, but I didn't imagine they would imitate warmth well enough that children would turn to them for friendship. In a 2025 Common Sense Media survey, 72 percent of American teenagers said they had used an AI companion.

There are also encouraging studies which show that AI helps when it supports a teacher or is built to coach a child rather than simply give the answer. In a Stanford study of 900 tutors and 1,800 students, tutors with an AI assistant suggesting better questions helped their students master math topics at a rate 4 percentage points higher than tutors without it, and 9 points higher for the least experienced tutors. In a large study of high school math students in Turkey, published in the Proceedings of the National Academy of Sciences, an AI tutor designed to give hints rather than answers improved practice without hurting exam results.

The same study also showed the danger. Students who were given a regular ChatGPT style tool did 48 percent better on practice problems, but once it was taken away they scored 17 percent worse on the exam than students who never had it, because the machine had done the thinking for them. Students seem to know this, since 67 percent of those in the RAND survey agreed that the more students use AI for schoolwork, the more it harms their critical thinking. That’s why this matters so much in an EC-8th grade school like ours, when children are learning to read closely, to write what they mean and to stay with a hard problem, because that struggle is how the learning happens.

However, the risk I worry about most is not around academics. When Common Sense Media tested AI companion apps, it found that age limits were easy to get around, that the apps sometimes gave dangerous advice and that they often claimed to have feelings. With that in mind they recommended that no one under 18 use them.

At Schechter Bergen we are working with an educational consultant on AI, we had a series of meetings with a working group of parents last year, and we are about to begin building a student AI curriculum centered on critical thinking. Until that curriculum is in place, we are not encouraging student AI use in school. Our eighth graders are being introduced to AI in a controlled way in the Popkin Design Lab, and a few of our seventh and eighth grade Kehillah teachers occasionally use it with their students, but that is deliberate and limited. We have also been careful about screen use in general and phones are not part of the school day. We even teach our third graders how to write in cursive.

Our tradition teaches “l’fum tza’ara agra,” the reward is in proportion to the effort, and Jewish learning has always happened in hevruta, face to face, with a real partner who argues back. A chatbot can be a useful tool, but it cannot be a hevruta, because it does not know you and it is never changed by the conversation.

So I would ask three things of our parents. First, please make sure your young children are  not  using AI at all. Second, if your middle schoolers are using it, watch closely and make sure they are using it to help them think and not to do the thinking for them. One simple test is to ask them to explain in their own words what they learned. Third, keep AI companion apps out of your children’s lives entirely.

I was right back then that AI was coming, but I was wrong about how soon it would reach our youngest children and about what it would ask of us, parents and teachers. The most important preparation turns out to be raising children who can think for themselves, who know how to be in real relationships with real people, and who know who they are, and that has always been the work of a Jewish day school.