THE ESSENTIALS IN 30 SECONDS

  • The gap in AI use between the highest- and lowest-income parents grew from 15 to 21 percentage points over 15 months, reaching 84% versus 63% in 2026, according to RAND.
  • In spring 2026, 57% of teens from families earning $150,000 or more used AI for homework, against 43% in families under $50,000; a year earlier both stood at 39%.
  • Teens' sense that their parents support AI use was more closely linked to their own use than any other factor studied except age.

A report published by RAND on 12 November suggests that family income is starting to shape how American teenagers use artificial intelligence for school, largely through their parents. The study, by Heather L. Schwartz and Melissa Kay Diliberti and sponsored by the Gates Foundation, draws on nationally representative surveys of about 1,000 parents and 1,800 teens aged 12 to 18, conducted in 2025 and again in May 2026.

Parents set the tone

Overall, the share of parents using AI for work or personal purposes rose from 54% to 72% between the two survey waves, RAND says. Growth was fastest among the best-off households: over 15 months, the gap between parents in the highest- and lowest-income brackets widened from 15 to 21 percentage points. In 2026, 84% of parents in families earning $150,000 or more used AI, compared with 63% in families earning under $50,000.

The same divide appears in what parents tell their children. According to RAND, 67% of higher-income parents encourage AI use for schoolwork, against 45% of lower-income parents. Most favour limited uses such as brainstorming or checking work, but 17% of the highest-income parents said they encourage their children to use AI wherever school rules allow.

A homework gap that did not exist a year ago

Among teens, AI use for homework was identical across income groups in spring 2025, at 39%. By spring 2026 it had reached 57% in families earning $150,000 or more and 43% in families under $50,000.

In a subset of 533 teens matched with their parents, perceived parental encouragement was the strongest predictor of use after age, ahead even of family income. Among teens who felt encouraged, 81% used AI for homework, versus 38% of those who felt discouraged.

What the authors recommend

For the researchers, what matters is how teens use AI, since earlier research suggests unrestricted use on a first attempt at schoolwork can undermine learning. They urge districts to block general-purpose chatbots on school devices during those first attempts, ask parents to discuss with teens the difference between AI doing the work and AI helping them think, and call on developers to build guardrails rather than outright blocks.

RAND says it is unclear whether this gap will widen or narrow existing achievement gaps. The findings rest on self-reported data and show associations, not causes.