Traditional learning techniques such as note-taking remain critical for students’ reading comprehension and retention, a new study by Cambridge University Press & Assessment and Microsoft Research has found.

The study also indicates that large language models (LLMs), such as ChatGPT and Microsoft Copilot, could support students in clarifying, exploring, and contextualising learning material.

Dr Jake Hofman, Senior Principal Researcher for Microsoft Research, says: “Rather than viewing traditional learning techniques, like note-taking, and newer generative-AI approaches as competing alternatives, we should view them as complementary.”

Although note-taking emerged as the more effective method for learning information, Cambridge Senior Researcher Dr Martina Kuvalja, who co-authored the study, explains: “Students found the AI chatbots more enjoyable and most said they would use the LLM method in the future.”

She adds that “being motivated to learn is half the battle,” but that “sometimes people want something that’s not good for them – we all love ice- cream, but we can’t eat it all the time!”

The study involved 405 students aged 14-15 attending seven different schools across England. Participants were asked to study texts on topics from the UK’s national history curriculum, including apartheid in South Africa and the Cuban missile crisis.

The students were divided into two groups. The first group studied one text with an LLM (ChatGPT 3.5 Turbo), and another text by taking notes. The second group also used the LLM for one text, but for the other text combined LLM use with note taking.

All students were given a brief tutorial before using the LLM and were allowed to use the tool however they wished.

Three days later, the students were asked questions on the two texts without prior notice to determine how well they had understood and remembered the information.

The results indicate that note-taking, whether used alone or alongside an LLM, leads to better understanding and retention of new information than using an LLM by itself.

Dr Martina Kulvaja explains: “No pain, no gain – if you make your own notes, you’re probably going to remember what you’ve learned better than if someone – or something – summarises it for you.”

However, students enjoyed using the LLM to explore relevant topics beyond the text. Dr Jake Hofman reflects: “I was struck by how many students used the LLM to try to deepen their understanding.”

The study’s findings could inform how LLMs are used in education. They suggest that students should receive guidance on how to use LLMs constructively, and that teachers could draw on students’ LLM interactions to identify where additional support is needed.

The study, published in the journal ‘Computers & Education’, is among the first randomised classroom experiments to examine how LLMs influence students’ reading comprehension and retention.

The National Literacy Trust found that, in 2025, 60.9% of young people aged 13 to 18 used generative AI to help with their homework.

The researchers plan to investigate further how different prompts, training and classroom designs affect learning when students use AI alongside traditional study techniques.