People have started asking language teachers a very strange question.
“Are you worried AI is going to take your job?”
I understand why they ask.
A machine can explain the Present Perfect. It can generate 10 gap-fill exercises in 3 seconds, give feedback on a paragraph, rewrite a sentence, produce a dialogue, simplify a text, translate a message, make a quiz, and probably design a role-play about ordering coffee in Paris before I finish this sentence.
So yes … I understand the question.
But I keep coming back to something Matt Coss said when we had him on the Teacher Talking Time podcast.
People ask him this constantly and his answer was sharp:
“No, but I’m worried that you would ask me that question because it tells me a lot about what you think my job is.”
And ever since then, that line has stayed with me,
You can hear the full conversation here.
What is interesting is that the AI question reveals the job description people have in their heads.
And sometimes, honestly, the job description is tiny.
Teach vocabulary. Explain grammar. Correct mistakes. Give practice. Test the forms.
That version of language teaching has always been easy to describe.
But why? Because it fits into a syllabus grid beautifully.
Week 1: present simple. Week 2: present continuous. Week 3: past simple. Sprinkle in some vocabulary topics, add a speaking activity near the end, call it communication.
I know that version because I taught inside it for years.
Early in my career, I could spend hours preparing a lesson that looked perfect on paper. The aim was clear. The target language was boxed in. The stages made sense. Students moved from explanation to controlled practice to freer production, and I could point to the board at the end and say: “See? That’s what we learned today.”
It felt safe.
But it also made teaching look like delivery.
And delivery is the easiest part of the job for AI to replicate.
Language education is pushing AI down our throats with chatbots.
A chatbot can produce an explanation of the second conditional.
A teacher sits in a room with 18 tired learners on a Wednesday morning and notices that the task has collapsed because the information gap wasn’t real, the instructions were too heavy, the stronger students have taken over, and the quieter ones are performing agreement rather than making meaning.
A chatbot can generate a list of phrasal verbs.
A teacher hears a learner reach for an idea, stumble into a form they almost have, and decides in that moment whether to recast, pause, board the phrase, let the interaction run, or save it for later because the communicative thread is still alive.
A chatbot can make a rubric.
A teacher looks at an entire program and asks whether students are actually leaving with the communicative abilities the program claims to build.
And Matt pushed this point hard in the episode.
If all we’re doing is vocabulary and grammar, AI makes our professional case look very thin.
Then he asked the better question: what do we actually do, and can we explain it to the people who make decisions about language programs?
That’s the part worth sitting with.
Many of us can defend our teaching beautifully to other language teachers. We can talk about input, interaction, output, focus on form, task design, assessment, learner needs, motivation, identity, belonging.
Then someone outside the field asks what students actually get from a language class, and we shrink the answer back down to grammar and vocabulary. Because those are the parts people recognize.
I see this in teacher education all the time.
Teachers do rich, complex work. They build relationships. They create conditions for learners to try things before they feel ready. They help students notice gaps in their own language. They design tasks where learners have to actually listen to each other because the outcome depends on it. They make a hundred small decisions in a lesson that never appear on any lesson plan.
Then, when they describe their work, they say: “We covered modals.”
The profession has long explained itself through the easiest parts to count. How many words, how many grammar points, how many units, how many levels.
Tasks make a different kind of claim.
A task says learners need opportunities to use language for a real purpose, with a genuine gap, with their own resources, and with an outcome that gives the communication some weight. Ellis has been clear about those criteria. Willis gave us a framework that takes classroom work past presentation and practice. Long kept pulling us back to needs, target tasks, and the real-world things people actually need to do through language.
And Samuda, in a way I find especially useful, reminds us that the task on paper is only the beginning. The task changes when it enters the room. The teacher shapes it, stretches it, rescues it, or sometimes accidentally strips away the very thing that made it a task in the first place.
That’s why this work is so hard.
And why it’s so human.
The strongest argument for language teachers in the age of AI is the full complexity of what we do when we take language learning seriously.
We design the conditions. We read the room. We listen for the gap. We decide when language needs attention. We manage the tension between meaning and form. We help learners build a relationship with uncertainty, because using another language almost always involves saying something before it feels fully ready.
AI can help with pieces of that. I use it. I think teachers should understand it.
But the current panic also gives us a mirror. And the mirror is uncomfortable because it shows how small language teaching can look when we describe it badly, design it narrowly, or reduce it to content delivery with a human face.
I keep coming back to that line from Matt.
The question tells me what you think my job is.
Maybe the work now is to answer with more precision.
My job is to create conditions in which people learn to use another language to do things that matter, with other people, in real time, under real constraints.
Try fitting that into a grammar box.
The full episode is a gem and we hope you find the time to listen!
If you like this post, then this is exactly the kind of work we dig into in TBL4T: task design, focus on form, classroom decision-making, assessment, and the messy middle where good intentions either become classroom practice or quietly slide back into old habits.
The September cohort opens September 19. Details here: 👉forms.gle/ndocoSb96zoSXx4JA