Using an AI Meeting Summary in a Conversation Class

If you’re a Zoom teacher, you may have noticed a new function of late: the ability to receive an AI summary of your meeting. I came across this before a recent conversation class with a group of adult learners (which just happened to be on the topic of AI). Intrigued, and having asked for everyone’s permission, I toggled the setting. Now, I’m not a Zoom spokesperson, but it’s no exaggeration to say that we were all blown away by the results; and, once the initial wonder had died down, I started to think about the possible uses for language learning.

Text by: Matt Prior

I’ll come on to those uses soon, but first, for anyone who hasn’t yet experimented with the function, what is an AI summary exactly? Well, a few minutes after the conversation class had finished, an email arrived in my inbox with a computer-generated review of the session: broken into relevant sub-sections and written in easy-to-read prose, it not only detailed what had been said and by whom, but more than that, it reported the information in a manner that offered reflection upon the learners’ attitudes towards what had been said. I exemplify below:

‘Maria expressed her concerns about the future of AI and the potential threats it poses, including the creation and spread of fake news and the possibility of AI making mistakes. She emphasized the importance of humans maintaining control over AI to mitigate these risks, and highlighted the benefits of AI in areas such as medical imaging. Maria concluded that while there are risks associated with AI, it has the potential to improve our world, and humans should strive to keep control of AI to maximize these benefits.’

I’ll say it again: we were all blown away.

How to take advantage of this technology in the online classroom, then?

It obviously makes for excellent reading material. I decided to forward the summary to the learners after each subsequent session and their reactions are always evidence of the extent to which they engage with the text. It’s highly personalised, and thus highly motivational. How often can a teacher say that about reading material? It helps them consolidate the conversation and, moreover, helps them upgrade their language. Maria, for example, did express the opinions in the paragraph above, but she didn’t express them in those exact words: strive, for instance, was unknown. How better to expose learners to new lexical items, which they can either look up in a dictionary or bring along to the next session?

AI large language models (LLMs) are prone to error, and that’s no different for the one Zoom employs here. Indeed, there are occasions when the summary isn’t quite accurate, or it assigns a comment to the wrong speaker. Why not turn this into another reading task? If a learner spots an error, they can either write it in the chat (in the case that the group has one, of course) or make a note for when they next meet. Additionally, this task has the bonus of training the learners to think critically whenever they are faced with computer-generated text.

I can think of many more typical reading and writing tasks that the summary could be used for (removing the sub-section headings, jumbling the paragraphs and asking the learners to unjumble them, for example). However, I think that the summary may actually be most effectively used as extensive reading material would be: without too much post-reading scrutiny, thus avoiding the risk that it loses the aforesaid almost-unique appeal that it carries.

Finally, it’s not only those present who the summary benefits: I’ve also had positive remarks from those absent, claiming that they didn’t feel like they had completely missed the session.

AI isn’t just coming, it’s already here. Let’s embrace it in language learning. How might you use an AI meeting summary in your online lesson, or any other new AI tool for that matter? Perhaps ChatGPT has a few ideas…

ELT News

ELT News