It took me a while to parse this plot, because of the strange way that The Economist has labeled their coordinates in the plot. The way to read it is: the relationship between homework scores and exam scores in the non-chatbot group is linear, but the relationship in the chatbot group definitely isn't. [![Exam scores versus homework scores, for students who did and did not use an AI chabot.][0]][0] {style="max-width: 50%; align: center;"} [0]: assets/nonlinear_grades.webp The [Economist article][1]; [as archived without a paywall][2]; the [technical paper][3]. [Via][4]. [1]: https://www.economist.com/graphic-detail/2026/08/18/does-ai-stop-children-from-learning [2]: https://archive.is/IpzHi [3]: https://dx.doi.org/10.2139/ssrn.6868618 [4]: https://bsky.app/profile/henrymance.ft.com/post/3mtgvx7hxbs2h