Can AI learn to teach?

Dr Karl Nightingale, Reader in Bioscience Education

In the previous step, you thought about the likelihood of robots replacing teachers by participating in a poll on this topic. In this step, to complete this activity, you will hear an academic’s thoughts on the same subject.

So …. will robots replace teachers?

This is clearly a divisive question.

For some, using AI in the learning process is an obvious step in the development of the technology, whereas for others it conjures up images of a dystopian future. However, as with many emerging technologies, the role of AI in learning is nuanced.

In the video for this step, I discuss the skills and qualities of an ideal teacher and how this includes:

• a broad and detailed ‘subject expertise’
• a high level of ‘teaching expertise’, which combines insight on how to develop a topic for learners (eg this comes with experience) and ‘human’ inter-personal communication skills, including intuition, patience and persistence.

Of these, I can see a future where AI addresses both ‘subject expertise’ and some elements of ‘teaching expertise’. This seems particularly likely for topics where there are clear and fixed boundaries (eg the rules of chess), where there is a large pool of learners and lots of experience developing learners’ understanding. You can imagine how this MOOC might develop if we had a huge choice of videos, supporting texts and quizzes to test learners’ understanding, and a ‘Chatbot-type avatar’ to guide learners, make recommendations, etc.

However, I can see three areas where this model of ‘AI-assisted learning‘ will be tested:

1. The first is in topics where the boundaries of our knowledge are expanding and learners therefore need to understand the way the field is developing and have an overview of the discipline. This is typical of undergraduate courses, which will need expert teaching staff to design and deliver up-to-date curricula.
2. The second is where there are practical or ‘hands-on’ skills involved in the discipline. This is central to the sciences, clinical disciplines and law, but oral communication and presentation skills are important in all disciplines.
3. A third is that the model assumes learners are motivated and will stay motivated, even if they encounter difficulties. The ability to ‘work things out on your own’ takes learners with self-knowledge, persistence and maturity – we can’t expect this of children. This is why learning with peers in an interactive, supportive environment maximises the chance that learners will engage and ‘stay the course’. Creating an interpersonal, reflexive learning environment is where excellent human teachers excel and will be difficult to replace with AI tools.

So, to summarise, there seems to be a future for AI-assisted learning, particularly for motivated adult learners in specific, well-defined topics (eg short courses, continuing personal development etc). However, incorporating AI into classrooms will be more challenging. It seems unlikely that AI will be replacing teachers any time soon.

When you have completed this step, you will have heard an academic’s thoughts on whether AI can learn to teach. You will also have completed this activity, in which you explored ways that generative AI could enrich teaching and learning.

Now that you have completed this activity, don’t forget to add some notes to your reflective log, which you can download from Step 1.1.

In the next activity, you will examine some of the immediate and longer-term potentials of generative AI in student assessment, marking and feedback.

Join the conversation

What do you think will be taught by robots/AI in the future and why?