Introduction
If you are a teacher or teacher training reading this, I’m sure you’ve wondered about what the best policy for AI usage is in your learning context. Recent research from Oxford University Press (2026[A11.1]) revealed that even for experienced researchers it remains uncertain what acceptable use is and what not. Whilst how to use AI in principled ways for learning is an important discussion to have, at times I wonder if we are asking the right question. In this blog, we will explore how AI usage may affect the human being behind the learner and what this means for our teaching.
Using AI
I’m sure many of us have been there. You need to write a complaint email and instead of opening a blank file on the computer you open your favourite chatbot and prompt it in detail to get what you need. Simple, right? So while there is no doubt that AI usage can be extremely helpful for efficiency, efficiency in learning may not lead to deepening language learning nor personal growth. Producing a polished piece of writing does not necessarily mean a learner has developed their language or writing skills.
And whilst there may be nothing wrong with using AI for idea generation for a project or essay, for spell checking or finding references or providing tailored self-directed practice opportunities (all depending on your circumstances), the overreliance on AI may take away more than it adds. AI usage may affect learners’ critical thinking skills (Gerlich, 2025) and learners’ ability to think independently, which is of particular importance when working with learners who are still developing their identity and need to strengthen their written skills. However, there seems to very little information around about what the impact of AI usage may be on learners’ confidence, motivation, identity and sense of self when a tool can for example write more fluently and coherently then learner. Basically, when we focus primarily on how to use AI ethically and what AI can contribute to support their learning, I feel that we risk overlooking the process of learning and even more so what may happen to learners’ sense of self when AI become their so-called crutch. So, what may happen?
Autonomy – whose voice are we hearing?
Autonomy is about having a sense of choice and ownership (Scharle & Szabo, 2000). In language learning, this often refers to the ability for learners to make decisions about how they communicate, to express their own ideas, and gradually take greater responsibility for their own learning.
AI can appear to support autonomy because learners can ask questions, have conversations, get feedback and explore sticky grammar points independently. Yet, there is a difference between making choices about what to ask AI to produce and making the decisions ourselves. Imagine asking learners to write about a meaningful personal experience. I know from experience that before they have even started to think about this, they’ve already prompted AI to generate a text. So while they this text may be grammatically accurate and coherence, where are their own thoughts, memories and feelings? What happens when they repeatedly allow a tool to choose the ideas and exact words to express their ideas? Over time, this may result in learners becoming less confident in trusting their own judgement or expressing themselves in their own personal voice. Their work may sound fluent but their own voice becomes less visible and a lot less them. If you ask me, part of our job is not simply to help learners produce better English but help them find their voice through English. Oh and I forgot to mention that we will have to read 27 essays that are very much alike…
Suggestions:
The aim is not to prevent learners from using support, but to ensure that they remain the authors of their own thinking.
• Ask learners to generate their own ideas on paper (yes, you read the correctly…) before even opening their chatbot.
• Ask them to ask AI to provide multiple chunks/ phrases to express their ideas. Next in pairs/ groups get them to decide which suggestions genuinely reflect their own voice and why they prefer one to the other what they want to say.
Competence: feeling capable without becoming capable
Think about a learner who struggled to write a paragraph in English. They spent time searching for vocabulary, trying different sentence structures and correcting mistakes. Eventually, they produced something they were proud of. Now imagine a learner asking AI to write a paragraph. Yes, the task is completed quickly, and most likely the result is more accurate, more fluent and perhaps a lot more sophisticated. But what have they learnt, and what do they now believe about their own abilities?
Self-Determination Theory, developed by Ryan and Deci (2002), identifies competence -as well as autonomy- as one of our basic psychological needs. We need to experience ourselves as capable of meeting challenges and making progress. Now, this does not mean that everything must be easy, or efficient. In fact, tasks of an appropriate level of challenge, combined with support, help learners deepen their skills and grow their abilities. When AI removes that desirable difficulty and effort needed, learners may experience what we could call a competence mirage (critical skills blog, 2026[A12.1]): their written work looks good, but learner’s independent ability to create this has not necessarily developed.
There is another danger. As teachers we know that writing skills are secondary biological skills and take hard work to strengthen, it takes time to write well and that is totally normal. However, when learners constantly compare their own developing English writing with the fluent output of AI, they may begin to see the difference in production time and sophistication as evidence that they are not good enough. Instead of thinking, ‘I cannot do this yet’, they may start thinking, ‘I will never be able to do this as well’ which may weaken their motivation. And sadly, it is true that they will never be able to write as well as AI. I mean, can you name anyone who can produce written pieces of text with accuracy at the speed of AI? Not that I know of!
Suggestions:
• Preserve opportunities for learners to experience genuine progress. Provide paper notes books (I know…) for some fluency writing tasks for example. They write 3 to 5 minutes without stopping 2 or 3 times a week. They make mistakes, they can revise their work and over the term they can recognise how much better they’ve gotten.
• Useful questions to discuss and take notes on before opening their favourite chatbot is: Could you do this without AI? If not yet, what would you need to learn to do it next time? After completing the task, ask them to reflect and identify one thing you could have done yourself and identify one area to work on independently to get a step closer to working on this more independently in future.
Relatedness: technology cannot replace human connection
Language learning is fundamentally relational. We learn to communicate by socialising and interacting with other people, interpreting meaning, responding to different perspectives and learning how our words affect others. Not only by studying words and grammar. AI can provide a low-pressure space in which learners can practise without worrying about making mistakes in front of others which can be particularly helpful for learners who lack confidence. However, we must distinguish between receiving a response and experiencing a human connection.
If learners increasingly turn to AI for “conversations” instead of people, they may miss opportunities to develop their social and emotional skills that meaningful human communication requires. Empathy, for example, involves more than producing polite or sympathetic words. It requires listening attentively, withholding judgement and trying to understand another person’s perspective before responding thoughtfully with curiosity and kindness. These are skills we only truly develop through putting interactions into practice. Personally, I believe that in ELT this matters enormously. We want learners to become people who can communicate across cultural differences, negotiate misunderstandings appropriately, express disagreement respectfully and build respectful relationships. A chatbot, while it may generate a supportive and accurate message, cannot develop these human abilities in context.
Suggestion:
• Our classroom must therefore remain a place where learners can experience various interactions patterns; pair work and group work so they talk to one another, listen to different viewpoints, collaborate and experience what it means to be understood by another person.
Identity: who are our learners becoming?
Perhaps the most important area of concern is how AI affects learners’ sense of self as a result of all the areas affected above. What messages are learners receiving about themselves through their use of AI?
Our sense of self develops partly through experience and partly through the messages we receive from others (Mercer, 2011). When teachers respond to learners’ ideas, acknowledge their contributions and give them opportunities to succeed, we help them develop beliefs about who they are and what they can do. The reverse, however, is also possible. If learners repeatedly receive the message that their own writing is inferior compared to AI, their ideas need improving by AI or their independent efforts are so slow compared to AI, they may begin to see themselves as less capable. This is especially important for younger learners, teenagers and adolescents who are still developing their identities, beliefs and sense of competence. But adult learners are not immune to lacking confidence!
Anyone learning a anything will most likely question their abilities when faced with a tool that appears to perform effortlessly what they find difficult. (I still have nightmares from trying to sow stitches in a straight line to then discover how easily and accurately a sewing machine could do this!). In other words, we need to be careful about the messages we send, both explicitly and implicitly. If we praise only the polished final product, we may unintentionally communicate that accuracy and fluency matter more than their ideas creativity, personal expression, effort put in and progress made.
So, where does that leave us?
Protecting learners’ sense of self does not mean throwing out the baby with the bathwater, it means that we need to make deliberate choices about when, why and how we use it. Below are five practical ways that can help us keep the human-in-the-loop, the learner at the centre:
• Before introducing AI, give learners time to think, discuss, brainstorm or draft ideas independently. Ask them to record their initial ideas. This will help them compare these with AI suggestions and decide what, if anything, they want to use.
• Rather than asking AI to complete a task, encourage learners to use it for a specific purpose: brain storming ideas, suggesting useful collocations, asking AI questions to help develop a stronger argument.
• Build in activities that require learners to work collaboratively and listen and respond to others. Pair discussions, information gaps, peer feedback, debates, gallery walks and problem-solving tasks all give learners opportunities to practise English while practising their social and emotional skills and build confidence to interact face-to-face.
• Create a reflection slips (or you can use that paper notebook…) that forces learners to reflect on the process of writing, not just their final work. What did they find difficult in the process? What did they work out themselves? Which AI suggestions did they reject, and why? What can they now do that they could not do before?
• Invite learners to reflect, in pairs or in writing, on how using AI makes them feel. Does it help them feel more confident, or does it make them doubt their own abilities? Are they learning something new, or simply finishing faster? Do they feel proud of the result? Could they explain or reproduce what they have produced? These conversations should not be about shaming learners for using AI, rather they could help learners become more intentional about their use.
Putting the human before the technology
AI is changing the way we teach and learn languages and this is a fac[A13.1]t, as Erasmus (2026) also highlighted in his blog. But we need to be careful not to allow the discussion, about what AI to use when, to distract us from the purpose of education itself. Our role is not to help learners become efficient users of digital tools producing better texts. Our role is to prepare learners to use AI critically and responsibly and develop their English language skills, their confidence and relationship skills that allow them to communicate effectively in the world.
I strongly believe that this means protecting learners’ opportunities to learn alongside others, to think independently, to make mistakes, to discover personal strengths and develop an authentic voice. This means remembering that learning is not just about what learners can produce today, but also about who they are becoming. Because we do not just teach a language. We teach people. And whatever technology we use, it is the human behind the learners that must remain at the heart of learning and teaching and that is where we can make a difference!
References
Oxford University Press (2026) https://fdslive.oup.com/www.oup.com/Group_comms/pdf/Disclosing_AI_use_in_academic_research.pdf
Ryan, R. M. & Deci, E. L. (2002). Overview of self-determination theory: An organismic-dialectical perspective. In E. L. Deci & R. M. Ryan (Eds.), Handbook of self-determination research (pp. 3–33). University of Rochester Press.
Scharle, A. and Szabo, A. (2000). Learner Autonomy A Guide to Developing Learner Responsibility. Cambridge University Press, Cambridge.
Mercer, S. (2011). Towards an understanding of language learner self-concept. Springer.
Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies, 15(1), Article 6. https://doi.org/10.3390/soc15010006
Critical skills blog (2026). https://criticalskillsblog.com/2026/03/10/the-competence-miraget-how-artificial-intelligence-has-made-it-easier-than-ever-to-appear-competent-and-harder-than-ever-to-actually-be-it/
