Learning with podcasts: when they really help and why your own material makes the difference

In short: For understanding facts, listening is almost as good as reading. But a podcast alone does little. It works when it meets three conditions: It comes from your own study material, it targets what you do not yet know, and afterwards you have to retrieve something from memory. A free podcast about a subject meets, at most, the first condition.
Is listening as good as reading?
Almost. A meta-analysis of 46 studies with nearly 4,700 participants directly compared reading and listening (Clinton-Lisell, 2022). Overall, it found no reliable difference. The two were on a par for facts and literal comprehension.
There are two limitations. People who could read at their own pace performed slightly better. And for drawing inferences, where you have to work out connections yourself, reading was clearly ahead.
For your studying, this means: A podcast is well suited to getting familiar with material and reviewing it. For deep understanding and application, you also need exercises that require you to think for yourself.
Listening alone is not enough
A study from Göttingen demonstrates this particularly clearly (Raupach and colleagues, 2015). Before each of ten lectures, medical students received a podcast and a quiz question to answer.
- Students who used both the podcast and the quiz question performed better on all ten questions in an unannounced test at the end of the module. Two months later, the advantage remained for seven out of ten questions.
- Podcast use explained the test results far more strongly than lecture attendance.
- Students who had watched the podcast but had not answered the question did no better than students who had used neither.
So the difference was not the podcast, but retrieval afterwards. This fits one of the best-supported findings in learning research: People who retrieve knowledge from memory retain it longer than those who simply read or hear it again.
Why your own material works better
So far, there are only a few studies on AI podcasts, and most are small. But they point in the same direction.
Personalised beats generic. In a study by Google and Drexel University involving 180 students, participants studied a textbook chapter as text, as a generic AI podcast, or as a podcast tailored to their field of study and interests (Do and colleagues, 2025). In philosophy, students using the personalised podcast scored 7.3 out of 10, compared with 5.9 for the generic podcast and 6.0 for the textbook. In the other two subjects, the advantage was smaller or inconclusive. One important finding: The generic podcast was no better than the textbook in any subject.
Integrated beats standalone. Google Research turned a textbook chapter into a learning environment with several formats, including an audio lesson, quiz questions, and diagrams (Heldreth and colleagues, 2025). Three to seven days later, learners remembered more than the group that only had the textbook: an average of 4.7 rather than 4.0 out of 6 points. The sample was small, with 60 school students, and Google conducted the study itself.
Questions during the podcast do not automatically help. In another study, the AI podcast regularly paused and asked for a brief reflection (Menon and colleagues, 2025). This did not improve learning outcomes, and participants liked the podcast less. Retrieval therefore belongs not in the middle of the podcast, but afterwards, as a separate exercise.
What nobody has studied yet: There is still no study directly comparing a podcast made from your own material, together with a study plan and exercises, against a generic free podcast. What we can say: Each individual component is supported by evidence. Personalisation helps, retrieval after listening helps, and spaced repetition helps. A podcast on its own, by contrast, is no better than a textbook.
Where and when podcasts fit
In a nursing degree programme in the United States, NotebookLM podcasts were used for exam topics (Bester and Zeigler, 2026). More than half of the students surveyed listened on their way to university. Two-thirds listened to most of the episodes. However, the instructors had to check every episode for factual errors beforehand. Whether the students achieved better grades as a result was not measured.
Podcasts therefore fit into the times when you are not at your desk: on the train, while exercising, while cooking, on your way home. They do not replace work at your desk, but they turn otherwise lost time into review time.
How to study with podcasts
- From your material, not from the internet. The podcast should cover what will appear in your exam, using the terminology from your lecture.
- Short and focused on one topic. Ten minutes on one topic is better than an hour on an entire chapter.
- Gaps first. A podcast is most useful when it covers something you do not yet know confidently.
- Retrieve afterwards. Do a few flashcards or a round of quiz questions immediately after listening. Without this step, little sticks.
- Review according to a plan. Listen to the podcast again after a few days, and use it as a recap shortly before the exam.
- Check what the AI says. Even good AI podcasts can make mistakes, especially with numbers and technical terms. When in doubt, go by the slide.
A free podcast or Learngine?
You can now get a generic podcast about a subject for free, for example with Google's NotebookLM. That is good for an overview. For exam preparation, it lacks what research says matters.
| Generic AI podcast | Learngine |
|---|---|
| Does not know what you already know | Focuses on your weaknesses and the mistakes from your last practice exam |
| Knows neither your exam nor your learning objectives | Follows your learning objectives and exam date (“9 days to go”) |
| You have to upload and organise the material yourself | Your material is already there; one click is enough |
| Chapters without links to the study material | Chapters link to the topic and slide, followed immediately by exercises |
| Runs outside your study plan | Appears as a block in your study plan (“Commute to university: 12-minute recap”) |
How Learngine does it
- Two voices, one topic, ten minutes. Structured as an introduction, mechanism, application, and summary.
- From your slides and notes. The podcast uses your terminology and memory aids.
- Available offline. For the train, exercise, and the journey home.
- Learning continues afterwards. After listening, Lumi suggests flashcards or a quiz on the same topic, so that you also retrieve what you have heard.
The AI makes the podcast. You do the retrieval afterwards yourself, because that is precisely when learning happens.
Frequently asked questions
Can I study with podcasts instead of books?
For facts and an overview, yes: Listening is almost as good as reading for these. For connections and application, you also need exercises in which you think and write for yourself.
How long should a study podcast be?
Keep it fairly short. A podcast of around ten minutes on one topic fits into a train journey and can be followed immediately by a short exercise.
Is listening to podcasts while doing other things useful?
During routine activities such as walking or cooking, yes, as review. You learn new, difficult material better with your full attention. And in both cases, the same rule applies: Retrieve it once afterwards.
Are AI podcasts reliable?
Usually, but not always. It is worth checking numbers, formulas, and technical terms against the slide. A podcast made from your own material is easier to verify because you know the source.
Sources
- Bester, M. E. & Zeigler, K. (April 10, 2026): Implementing artificial intelligence-generated podcasts in undergraduate courses. Teaching and Learning in Nursing, 21(3), 308–312. Article
- Heldreth, C. et al. (September 2025, revised March 2026): An Experimental Evaluation of an AI-Powered Interactive Learning Platform. Google Research, arXiv:2509.18664. Preprint
- Menon, V., Cherney, A., Cloude, E. B., Zhang, L. & Do, T. D. (August 2025): Evaluating the Impact of LLM-guided Reflection on Learning Outcomes with Interactive AI-Generated Educational Podcasts. arXiv:2508.04787. Preprint
- Do, T. D., Bin Shafqat, U., Ling, E. & Sarda, N. (2025): PAIGE: Examining Learning Outcomes and Experiences with Personalized AI-Generated Educational Podcasts. CHI 2025. Article
- Clinton-Lisell, V. (2022): Listening Ears or Reading Eyes: A Meta-Analysis of Reading and Listening Comprehension Comparisons. Review of Educational Research, 92(4). Summary
- Raupach, T. et al. (2015): Moving Knowledge Acquisition From the Lecture Hall to the Student Home: A Prospective Intervention Study. Journal of Medical Internet Research, 17(9), e223. Article
Learngine. As of November 7, 2026.
More from Knowledge
All articles on study methods and guides for university and school