Can AI Replace Teachers?

Here’s an opinionated video about the use of AI in classrooms. Let me know what you think. :slight_smile:

Birbalsingh points to the recent history of educational technology as proof of its inadequacy. She cites examples from Norway, Sweden, Denmark, and Finland, where initiatives to provide a device for every child consistently led to declines in reading scores and concentration. Many of these countries are now reversing course and reintroducing physical textbooks.

I’ve also posted an AI summary below the video.


Summary

In this video, Katharine Birbalsingh, known as “Britain’s strictest headmistress,” critiques the recurring belief among tech leaders—specifically citing Elon Musk, Sal Khan, and Sam Altman—that technology and AI can replace or drastically improve upon human teachers.

She argues that while these leaders are correct in observing that children crave responsive, engaging experiences (like those found in video games), their proposed solution—individualized AI tutors and screen-based learning—fundamentally misunderstands human nature and the true purpose of education.

Key Arguments:

The Failures of “EdTech” Are Documented

Birbalsingh points to the recent history of educational technology as proof of its inadequacy. She cites examples from Norway, Sweden, Denmark, and Finland, where initiatives to provide a device for every child consistently led to declines in reading scores and concentration. Many of these countries are now reversing course and reintroducing physical textbooks.

A Cognitive Crisis

She references neuroscientist Dr. Jared Horvath, who argues that the introduction of digital technology in schools coincides with a widespread decline in cognitive development among Gen Z. Horvath notes that increased screen time in educational settings correlates directly with poorer academic performance across dozens of countries.

The Misunderstanding of Motivation

The core flaw in the “tech-utopia” vision, according to Birbalsingh, is the assumption that children simply need information delivered at their personalized level to succeed. She asserts that, realistically, children often prefer entertainment over hard work and will default to distractions (like YouTube) when given a screen. She argues that learning requires self-discipline and working through difficult concepts—things children rarely choose to do on their own without adult guidance.

Education is Character Formation

Birbalsingh strongly rejects the idea that education is merely “downloading data and algorithms,” as suggested by Elon Musk. Instead, she defines education as fundamentally about character formation. It requires holding children accountable, managing their motivation, and addressing their work ethic.

The Irreplaceable Role of the Teacher

Technology lacks the capacity for moral formation. Birbalsingh likens a teacher to a personal trainer or a parent—an adult who not only possesses knowledge but also provides discipline, praise, emotional connection, and accountability. Children work hard, in large part, because of their relationship with their teacher and their desire to make the adults in their lives proud. An interactive screen cannot replicate this essential dynamic.

The Solution

Rather than seeking a “silver bullet” in new technology, Birbalsingh advocates for:

  1. Clear Expectations: Telling children exactly what is required of them.

  2. Responsiveness: Reacting appropriately to a child’s behavior, offering celebration for success and correction for failure.

  3. Better Teachers, Not Better Tech: Acknowledging that the current crisis is one of adult authority, not a lack of technology.

  4. Prioritizing Relationships: Recognizing that the relationship between a child and an adult authority figure is the single biggest factor in successful education and transformation.


Related Discussions

2 Likes

For people not familiar with this person…

…and there is plenty more controversy around her, so you guys might want to read up on her first before commenting. I for one have no comment.

I believe not. The educative system is rigid. And a teacher has a lot of work not only teaching. But private teachers yes are in risk cause of the ai I guess.

1 Like

The Guardian doesn’t like her, but she does get top results in an inner city school. I’ve read part of her book, and I think educators and parents should read it even if they disagree with many things in it. Vigorous debate leads to better ideas.

I’ve been through a lot of “alternative education” experiments, and her book has many useful criticisms of some of the things I’ve experienced.

When Gutenberg developed moveable type printing, some thought we would no longer need faculty. After all, why have your local physics professor when you can learn directly from Newton by reading, “Principia”? My sense is that AI tutors can be amazing teaching aids but I think for young people to learn, they will always need to gather with one another and with educators. As an aside - I think educators also act as mentors and can change the trajectory of a young person’s life. I don’t think AI will ever be able to do that.

4 Likes

Hi. I think the big problem about it is that educational system failed about make the students think from themselves . They only wanted hard work. Exams from natural memory and difficult written works. The ai had showed it clearly. The ai replace easily the tasks of research and create written works. Maybe after 200 years the education system have to teach to think not only repeating the books of others. This can be an opportunity to improve.

2 Likes

From reading the summary, I have to agree that AI alone, for most children, will not be sufficient. It would probably work for children who are intensely academically curious (e.g., Terence Tao), but that’s the exception proving the rule.

From personal experience, between the covid lockdowns and AI, my 12 yo nephew, an otherwise reasonably intelligent person, does not read at his grade level. He simply hasn’t had to, and he’s much more interested in playing video games on a tablet. He needs to be in class, with a teacher, without his tablet.

2 Likes

Although I’m not a huge fan of the “strict headteacher” branding, I 100% agree with Birbalsingh’s take in this video. It was somewhat saddening to see the young Swedish student reacting to her first physical maths book, seeing what they had missed and what my oldest kid missed out on at her school for 4 years. The comment that “what might be harder is that now we have to read for ourselves” is right on the money… that effort is part of the learning process!

I like that she also aired the arguments from Elon Musk and Sal Khan and considered their merits (e.g. that a sense of playfulness and personalised learning are good things) before disassembling the fatal flaws in the tech-first approach.

Edit: Removed full stops and corrected spelling mistakes.

Films, fiction books, games and TV series had higher accuracy than me in predicting the future although as I had read they were slow. As the things told in films of this century has not happened in theme.

People had learned through self study but I think that most can be done by AI once it gets trained on the right and additional data as one trick can save time learning to solve questions of chapters perhaps even of questions that require concept clarity as in MCQs even random guessing has 25% chance. Current tricks work on same wording or format etc. I have not found a trick that can convert the wording and format of questions into the standard format that worked for me.

Humans might have some pathways to deal with that problem and the pathways that lead to teachers existing might have higher probablity of chosing by Lindy’s law that the longer someting had existed the higher the chances of it for continued existence.

I would like to see what pathway we are in. Some questions might give better returns on deep work than others while the rote work can be done by the students themselves. The idea that some questions give better returns on deep work was gained by me trying to extend a theory and getting a line leading to me applying the 80-20 rule to deep work.

Perhaps 20% of question types give 80% of returns on deep work and the same with depth as atleast half of the tests on life up to the age 30 has most units of tests at 3/5 depth such as questions.

Competitive exams may be included in it or it might be an exception. More rigorous asking of AI might tell. I had written some of the statistics in other palces that may not be on the websites.

“The professions that did not adapt may no longer exist” (orginal source uncertain) but teachers could use AI to compound and add to their adaption speed and the same could be with other professions. I think this is where certain professions are going.

I guess for the broad professions that consist of multiple professions perhaps hundreds there might be less probablity of getting exitinct as some will adapt with AI and may use AI to boost their adaption speed. I had heard that AI will replace workers but I have not heard a single reason why. I don’t think that people make such claims without a reason.

Previously there was reading that was 200-300 WPM but then came writing a few thousand years ago that was 20x slower then might have come practice that was slower as people could read all the books bht not apply it without practice but then came revsion because of the decay and now people spend their lives mastering domains and have the risk of losing it to forgetting if they don’t revise. I guess that more things nad constrians will come to keep work alive and people will find ways to speed them up as memory techniqies did to memorisation.

I guess in future work might be more freelance employment, with stacked ranking system for work where people are paid a baseline fee and they have to compete with many other people to be in the top 10% or other percentage bands to get paid accoridingly and people might be able to do that multiple times in a day to earn as a in cooperatives as I think that two firms do. They could make tiny improvements and they could all be combined and compouned with a weighting system out of a few dominant ones to give the final product that might be more polished than current implementations. This is from blending the vibes of what people and AI have said and written.

Humanity based on an original analysis I read on an article had a constant battle to discover new resources before the rate of compounding if humanity destroyed the old resources on the Earth and I believe humanity always succeeded except when large numbers of humans died off in parts of time. From the human near extinctions research.

I guess it might be even easier to compete with AI as it would create new resources and someone would have to get them.

For teachers it could be the number of students who are unable to learn some or most topics to teach them those topics successfully.

I am waiting for constraints and accelarators and grading systems that would come to balance the AI as there was speaking, reading, writing, practice and revision as exist today. I am waiting for the next mode althogh I do think that would be slower. Creating a new mode that is slower and gives higher benifits might be a strategy of humanity and they may had emerged at the same time at different places “as differnet cultures over time faced the same problems and found overlapping solutions” “that solved the same constraints”" according to ChatGPT and Gemini.

If those modes require capablities and activities that only humans can do now then we might have an edge and overtime the creation of them might speed up and some might become better doable by machines. In future there could be AI teachers or AI teachers guided by humans perhaps in centralised,hybrid or decentralised command centers.

As there are some things that only humans can have context of as privacy might be valuable such as the confidential data of companies as they do not know what these AI might do and humans might be able to provide the context as privacy is valuable.

There could be things like patents and copyright to protect the humans and there could be words like plagarism that had not existed before and certain sharing might become rare as people and the teachers don’t share other’s copyrighted matarials live on video and they do not do that especially without crediting them. I guess that in the future there would be corporate hirearchies with some people trying to reach high positions to become partners or other roles in cooperatives and/or other forms of organisations.

I had before written about the expansion of copyright and patents being succeeded by others. I hope that the successor tools are pragmatic or positive and that they coexist with the other two tools.

I think she’s correct in all the points she talks about, and conversley the tech guys appear to know nothing about education. Gamification is one of the worst things that have happened to education. As the variuos countries mentioned in the video found out.

Besides, is there really a need for more tech and personalized tutors anyway? What will students learn that super intelligent AI won’t do much much better and much much faster in any and all areas?

Education will become decoupled from work since Super Intelligent AI and robots will be able to do all jobs, and we wont really need to be educated. Will we?

1 Like

Hi. What happened in that countries with gamification? I think is a good thing if it can be used in a clever way not replacing teachers. Traditional ways of teaching were not good either

The role of a grade school teacher is so complex and deeply vital to the developing minds of students that to even suggest that AI could replace it shows a complete lack of familiarity, or at least understanding, of the whole industry much less the job itself.

AI will be able to fully replicate the job of an airline pilot or dental surgeon before it ever comes close to imitating the job of a good childhood educator, the person a child sees as much or even more than their own parents.

I meant the countries mentioned in the video going back to basics. Gamification is built on the premise that education and learning should always be ‘fun’ and if it isn’t, there’s something wrong.
I think this is wrong (the video showed some evidence). For example, there has to be cognitive ‘friction’ for deeper processing to take place (as well as intrinsic motivation). That is, there must be cognitive struggle. It’s known that reading skills decline and attentional focus declines when primarily using screens.
The video showed evidence of these types of declines when tech is primarily used. And people like Musk are scary in their simplistic understanding of education.
Even the act of writing notes rather than typing them on a laptop improves retention and focus. We need to use evidence to make these kinds of decisions, not what some famous people think.
I haven’t seen evidence that tech and AI make people smarter, have better memories, have greater motivation etc.
Even in college and universities, many students rely on AI to completely write papers, “answer” chemistry questions etc. Is such a student able to put thoughts to paper in an organized and sophisticated way? Actually understand the chemistry to reply to questions? It usually shows up on the final exam, when the have done very well in all of the term work, but get an “F” on the final.