In this final section, we first consider the effects generative AI can have on teachers’ and learners’ wellbeing. We then explore what research has shown to be effective strategies for improving teaching and learning. We consider how these strategies can be applied to AI.

• 26 Considering wellbeing and AI

• 27 Carrying out action research into AI

• 28 Developing learner autonomy with AI

• 29 Developing your teaching with AI

• 30 What does the future hold?

Considering wellbeing and AI

Fears about AI

In early 2023, I started to notice a trend at the English language teacher conferences I attended. The rise of generative AI, particularly in its first widely known incarnation as ChatGPT, seemed to elicit two main responses from teachers. Some teachers welcomed how generative AI was saving them time with lesson planning and admin-related work (see 7). Other teachers were much more circumspect, and some teachers were frankly unnerved. At the end of my own conference talks on AI in ELT, teachers frequently asked me two questions. The first question was whether I thought that AI would replace them. The second question was whether AI would be used by learners to cheat on their homework (see 21 for my response to that). Both questions reveal deep uncertainties about our role as teachers.

Concern, and even fear, are not uncommon responses to new technologies (see 26). Generative AI has been perceived as threatening, not just by teachers but by society as a whole. Much debate continues to take place among philosophers, scientists and politicians about just how much of an existential threat generative AI might pose to humankind. You are likely to be aware of these debates and you may have listened to some of them, oscillating between fear and hope depending on who was being interviewed. The truth is, that at the time of writing this book, we are unsure of what the long-term (and even the short-term) effects of generative AI will be on us, and this can provoke a feeling of unease.

In this section, we consider some of the key ethical/moral, philosophical and legal issues that underlie the deployment of AI. Teachers and learners need to develop an understanding of these to use AI responsibly.

• 16 Can AI support language learning?

• 17 What does AI mean for teachers?

• 18 How can we make AI fair?

• 19 How can we make AI accessible to all?

• 20 Who owns the data?

• 21 Does AI help learners cheat?

• 22 Whose content does AI use?

• 23 Who creates AI?

• 24 Can we control AI?

• 25 How can we become critical users of AI?

Can AI support language learning?

The research conundrum

AI is present in a range of different learning technologies, as we saw in Section B. So, when we ask whether AI can support language learning, we need to specify what AI technology we mean. Once we’ve specified which AI technology we are talking about, we can explore what research has to say about it. Almost immediately though, we run into problems. Research studies are underpinned by different theoretical frameworks, and they ask different research questions. They utilise different research methodologies, look at different groups of learners and teachers, and the research takes place in very different contexts. This makes it difficult to compare the results of research studies that look at the same AI technologies. What works well for you in your context with your learners, may not work well for me in my context with my learners.

One way that researchers can compare different research studies is to undertake a systematic review of research that has already been carried out. By reviewing a range of research evidence, a systematic review aims to draw conclusions about the effectiveness (or otherwise) of a certain approach or a certain tool. Researchers usually set specific criteria for deciding which research studies to include in their review. For example, they may decide to include only research carried out in the last few years, and which has been published in peer-reviewed journals. So, to try and answer the question, ‘Can AI support language learning?’, let's look at what research in the form of systematic reviews has to say about two of the AI tools we explored in Section B.

In this section we explore whether – and if so, how – AI can be used to support language learning. We also look at some of the more controversial uses of AI in learning, such as using emotion AI to measure learner engagement.

• 6 Learning a language with AI

• 7 Teaching with generative AI

• 8 Personalising content for learners

• 9 Practising English with chatbots

• 10 Learning with augmented reality

• 11 Learning with virtual reality

• 12 Understanding real-time learner engagement through AI

• 13 Understanding emotion in texts with AI

• 14 Developing writing skills with AI

• 15 Assessing learners with AI

Learning a language with AI

A few months after ChatGPT was launched, I gave an online conference talk to language teachers about AI. In the question-and-answer session at the end of the talk, one of the attendees asked whether I thought that AI might be a passing fad. My answer? Definitely not. Here's why. Although AI in the form of tools like ChatGPT may have felt like a new development when it first appeared, AI has underpinned language learning software since the early days of Computer Assisted Language Learning (see 1). More recently, learners began using language learning apps on their mobile devices. Many of these apps offer automated language learning activities underpinned by rule-based AI (see 2). For example, some apps specialise in a specific area of language learning, such as grammar or vocabulary, which lend themselves well to a rule-based approach. The advent of data-driven AI (see 2) meant that language learning tools and apps became more sophisticated – machine translation and speech-to-text apps are just two examples of how access to large sets of data training has meant that these apps have got better (i.e., more accurate) over time.

Generative AI as a tool to support learning

In short, AI is not a fad in language teaching – or, indeed, in any other field. The advent of generative AI has had, and will continue to have, an impact on language learning software and approaches in interesting ways. Let's look at some examples of how generative AI can be used to support and enhance what we do in the language classroom.

This first section aims to demystify artificial intelligence (AI) by explaining in a very basic sense what AI is, and how it works. We also consider how and why EdTech is prone to hype, and how we and our learners might resist this hype.

• 1 What is AI?

• 2 What is generative AI?

• 3 AI and language learning

• 4 AI and creativity

• 5 Technology and the hype cycle

What is AI?

Early AI in language teaching With the amount of hype and hysteria that surrounded the arrival of ChatGPT in late 2022, you’d be forgiven for thinking that AI is a completely new technology for teachers and learners. Not so. The earliest and simplest forms of AI in ELT can be traced back to the 1960s, when CALL (Computer Assisted Language Learning) emerged as an area of study. In these early days of AI, computers could be programmed to provide limited responses to prompts. Computers were large and expensive, and tended to be found in universities. The advent of the personal computer, however, meant that by the late 1980s and early 1990s computers began to appear in schools and in people's homes. Language learning software with simple gap-fill and text reconstruction activities became available. As computing power increased, and computers developed multimedia capabilities, other uses for AI in language learning emerged. This was the heyday of the CDROM. By the early 1990s, some language learning software began to integrate voice recognition to support pronunciation. Since then, AI has become more powerful, and technology – especially in the form of mobile devices – has become more ubiquitous.

Narrow versus artificial general intelligence

To understand where AI has come from and where it is going, it is useful to distinguish narrow (or weak) and strong AI – the latter is usually referred to as artificial general intelligence or AGI. Here an analogy may be helpful. Imagine a chair that is mass-produced in a factory.