# Will AI Replace My Child's Teacher? What the Evidence Says **Author:** Dan Fitzpatrick **Canonical URL:** https://parents.theaieducator.io/posts/will-ai-replace-my-childs-teacher **Published:** 2026-10-16T07:17:57.000Z **Updated:** 2026-10-16T07:18:00.116Z **Category:** The Future ## TL;DR No, not on the evidence we have. Studies show AI tutors can help when they are built to teach and a teacher steers them, and one large study shows a plain chatbot can leave students worse off. England's Department for Education says its AI tools should complement teachers, not replace them. The schools run on AI have no independent results yet, and school rules come first. ## Key takeaways - The strongest studies test purpose-built tutors with a teacher involved, not the chatbot on your child's phone. - A standard chatbot used as a crutch left math students worse off once it was removed; a version with teaching prompts largely avoided that harm. - AI-run private schools such as Alpha School have no independent outcome studies that Axios could cite. - Ask your school which teacher oversees any AI tool, who evaluated it, and what happens to your child's data. - Tonight, ask your child what a teacher noticed about them this week. ## Article No. The best evidence so far shows AI helping children learn specific things when it is built as a tutor and a teacher stays in charge of it. It does not show AI doing what a teacher does, and England's Department for Education says the same. The caveat: most studies are short and narrow, and their tools are not the chatbot on your child's phone. > **The verdict** > > **Will AI replace my child's teacher?** Not on the evidence we have. The tools arriving in schools are designed to sit beside a teacher, and the best-known schools trying to run without one have no independent results that Axios could cite. > > **Does the answer change with age?** Yes. Under 8, the question barely applies. From about 13, your child may start using AI tutors on their own, and that is where the design of the tool matters most. > > **One thing to do tonight:** ask your child to name one thing a teacher noticed about them this week. It tells you what a screen would have missed. ## What does the research actually show about AI tutors? It shows that a well-built AI tutor can help, that a badly built one can hurt, and that the adult in the room matters. Here are the three studies parents most often meet in headlines, with what each one did and did not test. **Harvard, physics, 2023.** A randomized trial of 194 students in an introductory physics course for life-science students at Harvard University. Each student learned one topic with a custom AI tutor at home and another in a normal active-learning class. The paper, published in [Scientific Reports in June 2025](https://pmc.ncbi.nlm.nih.gov/articles/PMC12179260/), found the AI group scored higher on the follow-up test (median 4.5 against 3.5) and spent a median of 49 minutes on the lesson. The authors are careful about the limits. The tutor took several months to build, was loaded with expert-written answers, covered one course at one university, and measured learning straight after the lesson, not weeks later. It was also not a generic chatbot. **Nigeria, English.** World Bank researchers ran a six-week after-school program in nine public secondary schools in Edo State, Nigeria. Students worked in pairs with a free version of Microsoft Copilot, using prompts designed to make them explain and practice rather than collect answers. Teachers introduced each topic, suggested the first prompts, and ran short reflection exercises. The researchers, writing on [VoxDev in March 2026](https://www.voxdev.org/topic/education/how-ai-tutors-improved-learning-nigeria), report a gain of about 0.3 standard deviations on an English assessment. They attribute it to the whole package of AI, teacher guidance and structured prompts, not to AI alone. They also note that the underlying study is an unpublished manuscript. **High school math.** A field experiment with nearly 1,000 high school math students, published in [PNAS in June 2025](https://doi.org/10.1073/pnas.2422633122). Students given a standard GPT-4 chat interface did better while they practiced. When the tool was taken away, they did worse than students who had never had it. The authors say that without guardrails, students attempt to use GPT-4 as a "crutch" during practice. A version with built-in teaching prompts largely avoided the harm. That is the evidence. Here is my reading of it: AI helps when a person has designed the learning around it, and it can quietly cost children when it simply hands over answers. Nothing in these three studies removes the teacher. In the Nigerian trial, the researchers treat the teacher as part of the package that produced the gain. ## What are governments and schools actually saying? The English government is building AI tutoring tools and says they are meant to add to teachers, not replace them. In a [press release on January 26, 2026](https://www.gov.uk/government/news/450000-disadvantaged-pupils-could-benefit-from-ai-tutoring-tools), the Department for Education said the tools would work by "complementing high-quality face-to-face teaching – not replacing it." The release says the tools are being built with teachers, will be trialed with secondary pupils, and will be available to schools by the end of 2027. It says they could support up to 450,000 children a year on free school meals. An earlier [DfE release from June 10, 2025](https://www.gov.uk/government/news/ai-revolution-to-give-teachers-more-time-with-pupils) states that AI "should be used to ensure learning remains teacher-led." Both are statements of policy intent, not evidence that it will work out that way in every classroom. If you are outside England, check what your own state, district or education ministry has published, because the position varies. Parents are not asking for a swap either. In a Common Sense Media survey of 1,100 teens ages 12 to 17 and, separately, 1,244 parents, fielded in November 2025 and [reported by Education Week](https://www.edweek.org/technology/teens-say-they-should-be-able-to-use-ai-to-complete-assignments-parents-disagree/2026/03), 52 percent of the parents said schools should teach students to use AI. That is a figure about teaching children to use the tools. The survey, as reported, does not ask whether AI should replace teachers. ## What about the schools that run on AI instead of teachers? They exist, they are private and expensive, and they have not yet produced independent evidence. Alpha School is the best known. [Axios reported in August 2026](https://www.axios.com/2026/08/02/alpha-schools-ai-expansion-50-campuses) that students spend no more than two hours a day on AI software for core subjects, with adults called "guides" in place of licensed teachers, and tuition of $45,000 to $75,000 a year. Axios cited no independent study of the outcomes; the performance claims come from the company. In a Chicago Tribune piece [carried by Governing in April 2026](https://www.governing.com/artificial-intelligence/can-ai-replace-teachers-chicago-school-tests-the-model), Stanford's Victor Lee said of Alpha's AI software, "Right now, the evidence just is not there." He also pointed out that a private school with high tuition likely draws students who would score well anyway. Alpha co-founder MacKenzie Price said in the same piece, "We're not replacing teachers. The role is just changing." I would not call that a verdict against the model. It is an unproven model, run by a company, for families who can pay, and the one thing it plainly keeps is an adult in the room. The open question is what that adult is trained to do. > **A tutor can explain the problem again at nine at night. A teacher notices the child who has stopped asking.** ## What I See in Schools The pattern I see most often is that AI reaches a school through the teacher first, not the student. It starts as a way to draft a worksheet, adapt a reading for a struggling reader, or speed up written feedback, and the teacher decides what is used. That is a very different thing from the child facing a screen alone. What a teacher does that a tutoring tool does not is mostly invisible in a study. A teacher knows that this child is quiet because of something at home, that that one will not ask for help in front of friends, and that the answer on the page is right for the wrong reason. A teacher is also the adult who notices when a child is not all right, and in a school that matters. These are my observations from working in and around schools, not findings from the studies above. So my answer has two halves. AI will change how some of the teaching work gets done, and some tutoring will move onto screens. The part of teaching that is a relationship has not been shown to move with it. ## What would change my answer? I would want to look harder if your child's school did any of the following. Each of these is a reason to ask questions, not proof of a problem. - Replaced a qualified teacher with a supervisor for a core subject without telling parents. - Introduced an AI tool for students with no explanation of who evaluated it. - Used a tool that gives answers by default and has no teacher-set limits. - Could not say which adult is responsible for knowing how your child is doing. ## What to ask the school If your child's school is bringing in an AI tool, these questions are fair, and a good school will welcome them. 1. Which subjects use it, and for how much of the week? 2. Who chose it, and what evidence did they look at? 3. Which teacher reviews what my child does with it? 4. What happens to my child's data? 5. Can I see or try it? You could say: "We're glad the school is trying this. Could you tell us which teacher oversees it and how you'll check it's helping?" School rules come first on all of this. Where your child's school or district has a policy, it overrides anything in a blog post, including this one. I cover the wider question of what schools are doing in [How Is AI Actually Used in Schools?](https://parents.theaieducator.io/posts/how-is-ai-used-in-schools) and the rules for opting out in [Can I Opt My Child Out of AI at School?](https://parents.theaieducator.io/posts/can-i-opt-my-child-out-of-ai-at-school). If your child is already using a tutoring app, [AI Tutors for Kids](https://parents.theaieducator.io/posts/ai-tutor-for-kids) covers which kind helps. ## What do I tell my child who asks if AI will take their teacher's job? Say something true that fits their age, and do not promise what nobody can promise. A short answer is usually enough. **Under 8.** "Computers can help with some things, like practicing sounds. Your teacher knows you and what you need. That part isn't going anywhere." **8 to 12.** "Some schools are trying out computer helpers. They're built to help teachers, not do their job. If you ever use one, tell me what it was like." **13 to 15.** "Nobody knows exactly how this will look in ten years. The studies so far say AI works best when a teacher is steering it. What have you tried, and did it help you learn or just finish?" **16 to 18.** "Treat AI as a very fast tutor that doesn't know you. Use it to test yourself, not to skip the thinking. When you can explain an answer without it, you've learned it." For teens who use a chatbot on their own, the lesson from the math study is the practical one. Used as a crutch, it feels helpful and costs them later. Used to quiz them, it can help. My own line for families is: let AI help with the doing, keep the thinking with your child. ## What can't anyone promise yet? Nobody can promise how this will look in five years, and I will not pretend otherwise. The strongest studies are short, measure small groups, and test purpose-built tools. The three above do not follow children for years. The tools will change, the policies will change, and the evidence will keep arriving. If your child has additional learning needs, the school's special education team is the right voice for what suits them, ahead of any tool's marketing. What stays steady is the question worth asking whenever a new tool appears: who is the adult responsible for this child's learning, and do they know the tool well enough to steer it? If you would like your child's school to talk to parents about this, it is the work I do with schools, and [it is worth sending them this page](https://theaieducator.io/ai-training-for-schools?utm_source=parents.theaieducator.io&utm_medium=referral&utm_campaign=will-ai-replace-my-childs-teacher). I also write a Sunday newsletter that keeps teachers and parents up to date without the panic; [parents are very welcome](https://theaieducator.io/?utm_source=parents.theaieducator.io&utm_medium=referral&utm_campaign=will-ai-replace-my-childs-teacher#newsletter). *Dan Fitzpatrick is a former teacher and assistant headteacher, founder of The AI Educator, and has trained more than 150,000 educators. [About Dan](https://theaieducator.io/about?utm_source=parents.theaieducator.io&utm_medium=referral&utm_campaign=will-ai-replace-my-childs-teacher).* ## Frequently Asked Questions ### Can my child's school replace teachers with AI? Rules on who may teach a class vary by country and state, and I have not checked every one. In England, the Department for Education says its AI tutoring tools are meant to complement face-to-face teaching, not replace it. Ask your school which adult is responsible for each class and check your local rules. ### Are AI tutors good for kids? They can be, when they are built to teach rather than hand over answers. In a field experiment with nearly 1,000 high school math students, a standard chat interface left students worse off once it was removed, while a version with teaching prompts largely avoided that harm. ### What is Alpha School, and does it use teachers? Alpha School is a private network that has students spend no more than two hours a day on AI software for core subjects, with adults called guides in place of licensed teachers. Axios reported in August 2026 that tuition runs from $45,000 to $75,000 a year and cited no independent study of its outcomes. ### What should I tell my child who is worried AI will take their teacher's job? Tell them something true for their age: AI can help with some jobs, schools are building it to support teachers, and nobody knows exactly how this will look in ten years. Then ask what they have tried and whether it helped them learn or just finish. ### Should I stop my teenager using AI tutoring tools? Not necessarily, but pay attention to how they use them. The math study suggests a tool used as a crutch costs students later, while one used to quiz them can help. Your school's rules come first, and its teachers are the right people to ask about a specific tool.