Could AI Disrupt College Lectures and Note-Taking?

“Generative AI is about to disrupt lectures and note-taking in profound ways,” writes Marc Watkins, an instructor who teaches rhetoric and composition at the University of Mississippi. And that might even amount to the educational technology understatement of the year.

Whether they teach in online education or on-campus programs, most faculty at American universities still aren’t aware of how AI technology is poised to disrupt how students take notes during their lectures and class discussion sessions. However, whether this technology will actually confer any competitive academic advantages on its users in terms of better grades remains to be seen.

An entirely new breed of artificial intelligence-driven note-taking applications has started to spread like wildfire on college campuses across America in 2024. Running on laptops, these apps combine four simultaneous functions never before available in a single software application: audio and video recording, transcription, and artificial intelligence-driven outlining. And so much is happening all at once in these apps that all their operations are challenging to describe.

Not only do these applications record video and audio of an instructor’s lecture, but they transcribe the lecture’s audio track simultaneously in real time and display the transcribed text in a window on the laptop’s screen. The transcription isn’t much different from output produced by applications like Microsoft’s Dragon NaturallySpeaking, the first continuous-speech dictation software invented 27 years ago by Dr. James Baker at Carnegie Mellon University. It’s what these new apps next do with that transcription that’s so remarkable.

These applications simultaneously feed that transcript through a broadband WiFi connection into an artificial intelligence platform, such as OpenAI’s GPT-4o model, which currently powers ChatGPT. Pairing transcription with AI purportedly enables these apps to summarize the transcript’s main points while organizing those points into an outline of the professor’s script.

Then the app displays the outline constructed by the AI platform—essentially a machine-generated replacement for the student’s lecture notes—in a window next to the transcript in real time. In other words, while watching and listening to the professor’s lecture, the student can follow along as the app automatically writes the transcript in one window and converts the transcript into outlined lecture notes in an adjacent window. And all this technical wizardry happens at the same time.

Because these apps’ windows displaying all this scrolling real-time output need to be seen to be believed, here are a few example promotional videos recently posted on social media. The videos demonstrate two of these applications—JotBot and Notedly.ai—live in action.

This first example shows the Notedly.ai app on a student’s laptop recording a professor’s live lecture in an auditorium, generating a transcript, and meanwhile converting that transcript into lecture notes organized as an outline:

Where was this my freshman year? (Notedly.ai)

This next example shows the JotBot app similarly writing lecture notes for a student, but this time in action during a smaller classroom discussion section:

I am never taking notes in class again (JotBot)

Our third example should be of particular interest to online education students. This clip demonstrates JotBot in action, creating notes for a professor’s lecture playing on a YouTube video. Incidentally, the video appears to present an older lecture on schizophrenia by one of Stanford University’s most famous current professors, Dr. Robert Sapolsky:

JotBot saved my grade

One might expect from these demonstrations that the apps would spark controversy, and that now appears to be developing all over college campuses across the nation.

As the University of Mississippi’s Watkins points out in his Substack essay, most college faculty still have no idea about this particular artificial intelligence use case, nor are they aware of ways in which the software developers are marketing these apps to their students. Many observers might assume that the allure of these applications for students could lie in these apps’ capability to give their users academic advantages over competing students by enhancing learning effectiveness. In theory, the apps could accomplish such a result by enabling students to pay closer attention to the lecturer’s presentation, while liberating them from paying attention to their thoughts and decisions required by effective note-taking.

But that’s not Watkins’ analysis at all.

“It’s all just now coming to the forefront,” said Watkins, who as the Washington Post reported holds a dual appointment as director of the Mississippi Artificial Intelligence Institute along with his role teaching online and hybrid courses in the Department of Writing and Rhetoric. He told Inside Higher Ed in June 2024 that “the real challenge is faculty don’t know about it and a lot of the developers are going to social media and selling directly to students, not as a learning aid but as a mechanism to [stop their] listening skills.” He elaborates further in his essay:

These tools are increasingly marketed on social media to students not as a learning aid, but as another method to disengage and check out of the learning process. . .The problem with AI apps marketed to the public, and especially students, is that they equate learning with time-saving and little else.

Students take notes for many reasons, and one important reason for many students is that the note-taking process forces them to listen closely to the lecture. After all, one can’t exactly write accurate, organized lecture notes while daydreaming.

But if learners agree with sales pitches from app developers that purchasing an AI note-taking app means they can shut off their listening skills while the bot writes their lecture notes for them, they’re choosing to pursue a potentially risky study strategy. That’s because recent research on learning, memory and retention clearly indicates that those students risk missing out on valuable learning opportunities. And the loss of those opportunities could certainly factor into lower course grades.

In 2014 Dr. Michael Friedman of the Harvard Initiative for Learning and Teaching (HILT) compiled a review of the research literature on note-taking. In that widely-cited paper titled “Notes on Note-Taking: Review of Research and Insights for Students and Instructors,” Dr. Friedman writes:

Many mental processes occur simultaneously during the act of note-taking. The learner has to pay attention to the instructor, understand the material, identify what is important to write down in their notes, and coordinate the physical writing or typing of their notes, all while usually under severe time pressure.

That sounds like a lot of hard work. So are the learning benefits from lecture note-taking really worth all this effort? It turns out that Dr. Friedman concludes that yes, that’s indeed what the research in fact indicates:

Critically, learning can occur during both the production and review of notes by allowing the learner to make connections between idea units and engage in deep processing of course content (Bohay, Blakely, Tamplin, & Radvansky, 2011; Piolat, Olive, & Kellogg, 2005). The act of note-taking also assists the learner in generating and semantically processing information (essentially, helps the learner think about course content in such a way to better understand it upon later review), in addition to facilitating and strengthening the internal connections between ideas (Kiewra et al., 1991).

Lastly—and of particular importance to instructors—note-taking can result in broader learning outcomes in addition to improving mastery within course content due to this generative processing and making connections between idea units, allowing students to apply their gained knowledge to novel contexts (Peper & Mayer, 1978).”

One can recognize right away from this key passage that an AI bot’s writing lecture notes short-circuits all of these cognitive learning processes. Specifically, the bot bypasses learning opportunities in the following four ways:

• First, the learner wouldn’t have to pay close attention to the instructor, as they would in order to take effective notes.
• Second, they wouldn’t make connections between idea units and engage in deep processing of course content, to the extent those effects are promoted by note production.
• Third, the student would receive no cognitive prompting from taking notes in generating and semantically processing information, nor any prompting with facilitating and strengthening the internal connections between any ideas.
• And fourth, the lack of note-taking forestalls broader learning outcomes, and doesn’t promote improved mastery within course content without the generative processing connections between idea units that note-taking stimulates.

It’s true that while reviewing a lecture outline produced by an AI bot—assuming the outline provides an accurate representation of the professor’s script—the learner could certainly make connections between idea units, and engage in deep processing of the course’s content. And that’s probably even more likely if they’re also yellow-highlighting and annotating the outline. But by skipping their note-taking in the first place and instead letting the AI bot write notes, that student would nevertheless end up sacrificing all the value provided by four other cognitive processes that promote learning.

Could that learner end up earning the equivalent course grade without the learning and retention promoted by these four absent cognitive processes? That possibility involves a risk assessment judgment call that the student would need to determine. Keep in mind that competing students who don’t use the AI bot—those taking lecture and class notes in conventional ways—are likely to benefit from added learning during lectures accruing from all four of those cognitive processes. And that added learning could very well end up reflected in competitive academic advantages.

In other words, the student who uses the AI bot to write lecture notes would forego all the learning from those four omitted processes. Other things equal, relying on the bot for lecture notes could place that student at a competitive academic disadvantage against all the other students who took advantage of the additional learning promoted by their taking lecture notes in conventional ways.