Sarah Marzen
  • Main
  • People
  • Contact
  • Google Scholar
  • Random ruminations
  • Research Program
  • Conferences, Workshops, and Working Groups
  • Teaching
  • Main
  • People
  • Contact
  • Google Scholar
  • Random ruminations
  • Research Program
  • Conferences, Workshops, and Working Groups
  • Teaching

Random musings

Stray thoughts on my research, related research, education research, and sweeping commentaries on entire fields
Picture

Large Language Models and passive learning

9/1/2026

0 Comments

 
This is not a particularly new observation, I don't think, amongst educators who have taught students recently. A lot of students nowadays are using Large Language Models to complete their homework, get through study guides, and even take practice midterms and practice finals.

When educators talk about this phenomenon, it is often believed that the students are maliciously cheating using AI. At my colleges, what would constitute cheating in any other generation is often not done with any particular sense of malice. Oftentimes, the students are naturally wonderful learners who really want to get what's going on in the course, and they want to use AI to supercharge themselves.

The problem is that AI does not supercharge student learning in general. One study looked at the difference between Chinese students' homework scores and exam scores. The students who used AI did quite badly on the test, while acing the homework. I severely doubt that everyone using the AI was simply copying and pasting.

I think the reason for the lack of real learning when using AI is very simple. AI is, unless you're in study mode, entirely passive. You ask a question, you get an answer, and you read and nod. A lot of work has been done on the difference between this kind of learning-- which is passive learning-- and active learning, in which you fight to understand the material by asking yourself questions about what's happening, doing practice problems, and essentially stumping yourself en route to a better understanding. Active learning can include techniques like think-pair-share, group work, or clicker questions. Active learning techniques are known to outperform passive learning techniques in terms of retention, while also being less enjoyable for students. Why? Who wants to exert energy? Nobody. We just really want the answer. But it's better for us to work for the answer.

What this boils down to is that it's nearly impossible to use AI in a way that supercharges learning. This does not mean that AI cannot be used for generating lesson plans or helping with research projects if you're having trouble coding within the allotted time, like I discuss with some students here. But it does mean that study mode is a must. Those questions are not just a nice idea that some people had when they wanted to help students; they are essential. One study that I can't find now showed that study mode saved students from losing retention of material while using AI.

I think when we talk about how to use AI to supercharge humanity, we have to remember that one of the most important things we do as humans is grow, and passive learning means that there is, essentially, no growth.
0 Comments



Leave a Reply.

    Author

    Write something about yourself. No need to be fancy, just an overview.

    Archives

    November 2026
    October 2026
    September 2026
    July 2026
    June 2026
    May 2026
    August 2025
    February 2025
    January 2025
    December 2024
    August 2024
    July 2024
    May 2024
    December 2023
    October 2023
    November 2022
    July 2022
    December 2021
    November 2021
    March 2021
    February 2021
    May 2020
    July 2019
    May 2017

    Categories

    All

    RSS Feed

Proudly powered by Weebly