TL;DR: A language school has four options with AI tutors: ignore them, compete on what they lack, use them inside your courses between lessons, or build or resell an AI product. For most small schools the third is the realistic one: AI practice that the teacher sets and reviews, described honestly on the website. A decision table, a cost worksheet and the rules for using AI tutors follow.
The owner of a small language school sees the same ad three times in a week: an AI tutor app that promises speaking practice any time, for a monthly price lower than two lessons. In the same month, two adult learners do not re-enrol. One says she will "try the app for a while". The teachers have opinions, and they do not agree.
Should a language school use AI tutors or compete with them? There is no single answer, but there is a way to decide. An AI tutor here means software that holds a conversation with a learner in the target language, corrects them and explains, without a teacher present. This article gives you four options, a decision table, a cost worksheet and the rules for using AI tutors well. It is part of our guide to language school marketing in the age of AI.
What does an AI tutor do well, and what does it not?
It gives unlimited practice at any hour, with instant correction and patient explanation. It does not place a learner at the right level, plan a course, notice a pattern of errors across a term, or hold anyone to turning up. Those gaps are what a school sells; the strengths are what a school can use.
| An AI tutor does well | An AI tutor does not do |
|---|---|
| Practice dialogues at any hour | Place a learner reliably at a CEFR level |
| Instant correction of written text | Track the same errors across weeks and adjust the plan |
| Patient explanation, repeated as often as needed | Prepare for a specific exam's tasks and timing as a teacher does |
| Vocabulary drills on any topic | Make a learner turn up on Tuesday |
| Practice in a low-stress setting | Notice when motivation drops, and react |
AI tutors also make mistakes. They can accept a wrong form, invent a rule or drift into a register no learner needs. A learner at A2 cannot tell. That is one more reason the teacher stays in the loop if the school uses them.
What do teachers think of AI tutors?
Teachers are divided, and many already use AI themselves. About half do not expect AI to teach without a teacher in the next decade, but a large share use AI tools for materials and practice. A school deciding on AI tutors should ask its own teachers first, because they will make any choice work or fail.
In a British Council survey of 1,348 English teachers in 118 countries, 24% said they used no AI tools (British Council, 2024). The most common uses were creating learning materials at 57%, helping learners practise English at 53% and creating lesson plans at 43% (ICEF Monitor, 2023).
The same survey asked whether AI will teach English without a teacher by 2035: 51% disagreed, 24% agreed and 26% were neutral (ICEF Monitor, 2023). That is teachers' opinion about the future, not a measured result, so use it to frame a staff discussion, not as a marketing claim.
The decision framework: four options
Ignore, compete, use inside your courses, or build or resell. Each fits a different school. The table shows when each one makes sense, what it costs, what can go wrong and what the website then says. Most small schools end up combining the second and third.
| Option | When it fits | What it costs | Main risk | What the website says |
|---|---|---|---|---|
| 1. Ignore | Your learners are mainly juniors or exam groups who need a teacher anyway | Nothing now | Adult learners leave for apps without saying why | Nothing about AI, which applicants notice |
| 2. Compete on what AI lacks | Your strength is speaking, groups, exam strategy | Rewriting pages; teacher time on videos | Sounding defensive | What the teacher adds, in checkable terms |
| 3. Use AI tutors inside courses | Learners need more practice between lessons | Licences; teacher time to set and review tasks | Unchecked AI errors; data handling | "AI practice between lessons, set and reviewed by your teacher" |
| 4. Build or resell an AI product | You have a large learner base and a technical partner | Development or reseller terms; support | A product that competes with your own courses | A separate product with its own honest description |
Option 2 is mostly a writing task; the page structure is in why pay for language lessons when there is ChatGPT. Option 3 is mostly an operations task, covered below. Option 4 is a different business and rarely the first step for a school with a handful of teachers.
Example: deciding with a cost worksheet
Put the option you are leaning towards into numbers before you decide. Two lines are enough: the extra cost per learner per term, and the teacher hours the option adds or saves. Compare them with the course fee and with what you know about why learners leave.
The formulas:
- Cost per learner per term = (licence cost per month × months) ÷ learners using it.
- Teacher time per term = (minutes per learner per week to set and review AI tasks × learners × weeks) ÷ 60.
Example (illustrative, not a real case; amounts in euros): a school with 60 adult learners considers option 3 for its B1 and B2 groups, 36 learners in all. A school licence costs 90 a month, for four months of a term: 360 ÷ 36 = 10 per learner per term, against a course fee of 240. Each teacher plans 5 minutes per learner per week to set a task and look at the transcript: 5 × 36 × 12 ÷ 60 = 36 teacher hours a term, which is the bigger cost. The owner decides to pilot with one B1 group of 8 for one term (8 hours of teacher time), to ask those learners at the end whether the practice helped, and to decide on the rest after that.
If you use AI tutors: the rules
The teacher sets the task, the teacher reviews the result, and the learner knows which is which. Keep learner data in tools the school has approved, check what the tool stores, and never describe the AI as a teacher. These rules protect learners and keep your marketing claims true.
- Teacher sets, teacher reviews. Each AI task links to the current lesson, and the teacher looks at a sample of transcripts each week.
- Approved tools only. Learners' names, levels and test results go only into tools the school has checked for how they store and use data.
- Clear labels. In the course and on the website, call it "AI practice", not "AI teacher" or "personal tutor".
- A way to flag errors. Learners can mark an AI answer they doubt, and the teacher answers in the next lesson.
- Records stay in one place. Attendance tracking and practice records should live in the school's own system, not only in the AI vendor's account.
Your school management system may already hold part of this. One vendor describes its software as covering teacher scheduling, attendance tracking, billing, a lead and student CRM, a student portal and payment reminders, with AI-drafted fee reminders approved by an admin (Teach 'n Go, 2026). Ask your own provider whether it connects to the AI practice tool you are considering, or at least exports records.
How do you describe AI tutors on your website?
Plainly, in one place, and in the same words everywhere: what the AI does, what the teacher does, and what the learner does. Applicants will compare your description with what an assistant or an app tells them, so it has to be specific and checkable, without promising faster results.
Applicants increasingly ask AI assistants before they ask you. In BrightLocal's 2026 survey, 45% of consumers had used ChatGPT or another generative AI tool for local recommendations, up from 6% the year before; the figures cover all sectors, not education (BrightLocal, 2026). A clear sentence on your course page is what an assistant can repeat.
Wording that works: "Between lessons you practise speaking with an AI tool for about 15 minutes, three times a week. Your teacher sets the topic and looks at your transcripts every week. The AI can make mistakes; mark anything that looks wrong, and we go through it in class." Wording to avoid: "Your personal AI teacher, 24/7", "learn twice as fast with AI", "AI-powered fluency".
For the course pages that carry this sentence, see the course description template; for how to show results from blended courses honestly, see showing learner results without promising fluency. The wider picture is in the language school marketing guide.
If you want one place for your school's decisions, tool rules and prompts, so the AI policy does not live in one person's head, we are building the Business Brain Starter Kit. Get notified when it launches.
FAQ
Will AI tutors replace language schools?
Not for learners who need a level placement, a plan, speaking with real people, exam strategy or someone who expects them each week. AI tutors give unlimited practice and instant correction, which some learners will choose instead of a course. A school that names what it adds, and uses AI practice inside its courses, gives those learners a reason to stay.
Should we let learners use AI tutors for homework?
Yes, if the teacher sets the task and looks at the results. AI practice works best as a structured part of the course: a topic linked to the lesson, a set length, and transcripts the teacher can check. Tell learners the AI can be wrong and give them a way to flag doubtful answers for the next lesson.
Can we call an AI tool a personal tutor in our marketing?
Avoid it. Calling software a tutor or teacher suggests a level of judgement and responsibility it does not have, and learners who believe it may be disappointed. Describe what it does in plain terms, such as AI speaking practice between lessons, and say clearly who sets the tasks and who reviews the results.
Sources
- Artificial intelligence and English language teaching: preparing for the future — British Council, 2024.
- Global survey says English teachers are both enthusiastic and concerned about AI — ICEF Monitor, 2023.
- Teach 'n Go: language school management software — Teach 'n Go (vendor description), accessed 2026.
- Local Consumer Review Survey 2026 — BrightLocal, 2026.