The search never ends for ways to get better grades.
Lately, sales of AI -powered education devices are picking up in the Chinese market, meaning parents are willing to spend good money on smart tablets or watches, which they believe will give their children a leg up.
Hello I'm Lai Ming and this is Roundtable.
Today with Li Yi and Steve Hatherly we find out whether AI education devices will rise to change the landscape of education and the related industries.
The concept of AI education devices, well it's rather complicated, consists of multiple devices like we mentioned earlier, smart tablets and watches, in what other forms do they come and what purposes do they serve?
Yes, as you said I think AI learning machine is just an umbrella term for a whole bunch of smart portable devices that are integrated with AI technology and also designed for educational purposes for students and the most common ones, as you said, are smart learning tablets and smart watches and in fact
I think learning machine is not something new here in China, I mean back in 1980s we've already witnessed a surge of learning devices here in China.
I had one. You had one really?
In the 80s, what did you have?
It was a Xueshizi, I couldn't recall the brand, my dad brought it back from Shenzhen, southern China.
At that time it can be super expensive actually to afford such learning machine and some typical well known brands could be Xiaobao Wang.
That's more like a gaming console.
Yeah and I know they're super popular among young kids in China back then and also I think these I mean those learning machines offer pretty primary educational functions that help you recite English words and also programming and typing could also be some advanced function.
I learned to type in Chinese characters using that device.
Yeah that is very popular and can you imagine in 2008 alone the sales of reading pens actually exceeded 2 million units here in China.
So that's a very popular usage of learning machines here in China and then we see I mean even more popular tablets and more digital equipment begin enter education sector here in China because you have more educational tablets with various educational apps and so people start to buy these learning machines
due to their relatively new functions and more abundant educational resources and now it's becoming AI.
Yeah and it's become a huge market here in China.
The market for educational devices in the country it's going to be worth they say 20 billion dollars in 2026.
That's according to a research firm by the name of Frost and Sullivan and that's just proof that yeah it started back in the 1980s but it's really never gone away.
And now it seems to be more and more prevalent and a lot of people are making videos about it online as well as of September 5th of this year videos on quote unquote learning devices that inspired more than 530 ,000 posts on social media platforms here in the country and it got views of over 3 .8 billion really
impressive numbers.
Videos are titled like this.
I spent 30 ,000 yuan to test out these five most popular AI learning devices myself.
You can find lots of videos online like that by the way 30 ,000 yuan is about 4 ,250 US dollars.
It's getting popular but I think AI learning machine is pretty niche market here in China because when you look at when was the first AI learning machine created actually every brand has their own answer they're all claiming they have invented the very first AI learning machine here in China but I think
generally it is in the 2010s that we see this rise of AI learning machines and of course this market has developed with advancements in AI technology.
So that's really making this AI learning machine sector pretty new.
But as Steve said it sounds pretty good here in China because data from scientific hub rental technology suggests that in the first half of this year the total sales of learning devices across all China's in China actually was over 2 .5 million.
That's in over 20 percent interest compared to that in 2023.
So that's why this learning machine has inspired discussion and also debates among not only parents but also the general public here in China.
So there's intention and demand for AI powered education devices or education assistant devices.
People are actually searching for them and hoping to learn more about them to help them decide what to buy and there are also statistics showing that people are wanting to spend good money on that.
But what is it people are looking for specifically AI powered devices?
I remember with the advent of smart phones and smart tablets one can download anything there's a saying there's an app for anything.
So it's been quite a few years when parents are quite used to downloading an application powered by extraterritorial training service providers.
Could be Yuan Fu Dao could be a shares in Chinese.
And why is it people couldn't just simply rely on these applications but instead are looking for an AI powered device.
I think to answer your question we can look at some of the major features that these AI learning machines are offering.
As you said they're pretty expensive actually because one AI learning machine could cost at least 1000 Yuan and I checked the most advanced type could be as expensive as 10 ,000 Yuan.
That's over one that's over 1400 US dollars.
So that can be more expensive than a smart phone.
Exactly and also I checked the… Well depending on the smart phone.
Yes and I think the best selling machines are typically at 2000 or 3000 Yuan but that's still quite a lot of money.
And when you look at a very advanced device that is super popular in the market produced by one major brand that is very good at producing AI language models.
And this model is sold at nearly 9000 Yuan that's over 1200 US dollars.
And according to its official advertisement it offered AI advancement courses for kids not only in primary school but also in kindergarten and middle schools.
And the subjects math, English, and Chinese.
And it can really offer 35 AI learning assistance features including reading pens, AI translation, AI homework correction, AI essay correction.
And you can even take a picture of say an essay and also some problem that you meet when you are learning and the machine could just offer you explanation.
And also it can help you make AI learning schedule.
A lot of the things I think should be something that teachers might be able to facilitate when students are in schools.
Why is it that parents are seeking similar services that their kids can access when they're in school?
Well you have a tutor at your fingertips I think is the answer to that question.
I was watching a TED talk and the man who was doing the TED talk his name was Sal Khan and he's the founder of a free online educational platform.
Khan Academy. Yeah exactly.
And he said with these tools, with these AI tools, you're giving your child an amazing tutor in their hands.
And he referenced and I thought this was quite interesting.
He talked about Benjamin Bloom's 1984 two -sigma problem.
Benjamin Bloom was an educational psychologist.
And without going too much into that study, Bloom's two -sigma problem refers to the educational phenomenon that the average student tutored one to one using mastery learning techniques performed two standard deviations better than students educated in a classroom environment.
Long story short, according to Benjamin Bloom and his study, private tutoring works.
And another point on tutoring I want to go to another country to illustrate this example.
Last year, officials in Korea made the move to remove so -called killer questions from the the Gao Kao, essentially in Korea.
And at the time the education minister, his name was Eju Ho, he said these notoriously difficult questions sometimes include material that isn't covered in public school curriculum.
And this lends an unfair advantage to students without access to private tutoring.
He said it was a personal choice for parents to send their kids to private tutors or to institutions.
Many feel forced to do that because of the intense competition to do well on the exam.
But for lower income families, and I am going to get to my point in just a second, for lower income families, and this is from the borganproject .org.
This is from 2021. An estimated 16 .5 % of poor households overspend on institutes in Korea.
They invest about 30 % of their income as opposed to the 5 % average among higher income households.
So these cram schools or these institutes, they demonstrate how parental employment impacts educational inequality in South Korea.
The average monthly spending on private education in the country was 434 ,000 won a month.
That's about 2 ,291 yuan.
And that's not cheap per month.
So to finish my point is that if you can put an AI tool into a child's hands to help them keep up with the other students who perhaps are doing a little bit better in math or in English, and they can use that tool outside of the classroom.
You're right, Limey.
Yeah, teachers are teaching in the classrooms, of course, but kids need extra help sometimes.
And if these AI tools can provide that assistance to families and make educational, to make education available to everyone and equal for everyone, then this has to be considered a good thing.
So in theory, one -on -one tutoring sessions sometimes can make up for what students aren't getting when they're in our classroom context, where one teacher might need to address the demands of several dozens of students.
That makes sense. But then my follow up question is, will the AI education devices make up for that shortfall?
Will they work or serve as substitutes for one -on -one tutoring?
I think to some extent, yes, because as Steve said, I mean, parents in South Korea, they're paying huge attention to children's education.
And I think pretty much a similar thing here in China, especially those parents in urban areas.
And they're more than willing to spend huge in investing in children's education.
Because when you look at, I mean, who are using, who are buying these expensive learning machines, actually, according to a latest survey released by iResearch in September, the majority of parents who have been responsible for their children's education and have purchased educational smart devices
in the past two years are mothers aged 30, between 31 to 45.
And most of them live in economically developed first tier or second tier cities.
And their household income typically between 10 ,000 to 50 ,000 UN.
That's about 1 ,400 to $7 ,100 US dollars.
And also, they primarily work in office jobs with around 70 % in management roles.
So when you look at the demographic of these, I mean, users of AI learning machines, first of all, these are parents of children who are willing to spend money and who pay attention to good education of their kids, and also who can afford this huge AI learning machines.
And most importantly, they might not already have time to provide this private tutor to their children on a daily basis.
So back to your question, is it really effective?
I think if, as a parent, if you are, if you do, I mean, care so much about your children's education, yet you don't already have time.
So when I already buy this AI learning machine, I assume that's one possible reason.
And two, don't forget, there are parents who can't really offer efficient tutoring to their children who might already have this knowledge to offer help.
So in this case, they can also buy this AI learning machines.
You described potential or existing clients or consumers for AI powered education devices.
And in your description, I think it's more like parents who are better off, comparatively speaking, who pay more attention to their kids' studies are more likely to do that investment.
I think in comparison, AI devices are not exactly serving the purpose of making education more equitable and more fair.
But it goes to show that there is plenty of market potential for AI education devices to go even further.
But you can't disagree.
Sorry to interrupt, but you can't disagree that it's a more affordable option.
I'm not saying that these devices still aren't cheap.
They're not cheap. And for lower income families, even investing in one of these AI devices is going to be really, really expensive.
But if you compare that to, what did I say about Korea?
$434 ,100, $2 ,200 yuan a month.
A month. So a one time purchase of a device versus a monthly charge of going to a private tutor or to an institution, at least we can say these devices are more affordable.
Mm -hmm. Per Li Yiyi's description, she mentioned better off families, parents who probably work office jobs, they're in a better position and with better educational background to be able to give the children the guidance that they need.
In comparison, parents who work menial jobs, who don't necessarily have had that as good of education background as the office workers, they probably don't have the knowledge to be able to teach their kids tutor their kids, even if they were able to afford the time.
So in that case, I think the AI education devices has a big market to grow even further.
But the question is, are they up to the job?
How good are they in providing that one on one or individualized, personalized service, tutoring service?
I think that's a very good question because when you look at whether those AI learning machines are being so effective in terms of improving children's performance, especially academically and also at school, that is a question mark.
Because some people are saying that this AI learning machine is still being too smart in a way that is too convenient and too efficient.
For example, we mentioned a feature that this AI learning machine is offering that children can just take pictures of, say, certain essays or certain problems that meet while they're studying.
And then the AI learning machine will just quickly offer you an answer.
And even before this rise of AI learning machines, there is so -called homework assistance apps that are so popular here in China, children can also, while they're doing their homework, if they don't really know the answer, they can just take the photo, take the picture, and upload a picture, and the machine
will tell you the answer.
So according to some of my friends who work in this education industry, they are not already so optimistic about the usage of these AI learning machines in the way that they don't really help cultivate the children's capability of studying and thinking and doing self -reflection because it's so quick.
It doesn't really allow you this process of digesting everything and then do the critical thinking by yourself.
This somehow reminds me of the debate between electronic dictionary and paper dictionary.
I'm sure, I mean, you must share the similar feeling.
I mean, well, because I'm a language learner and while we are studying language, of course, at the very first beginning, we use paper dictionary.
They're so heavy, they're so thick, they're not really so convenient to carry.
And then we have electronic dictionary, right?
They're so convenient and they're so quick and the capability, I mean, the capacity of words is so huge than paper dictionary.
But despite our teachers, do you recommend us to use paper dictionaries?
Because while you're using this paper dictionary, it really offers you the time to recite, to reflect and then you have better memory of those words, right?
So I think similar situation here in this AI learning machine, I don't really know.
No, with the digital dictionary, I think it makes sense because I think people, human beings, we remember by association and we don't necessarily just remember a concept like a car is a car.
But sometimes if you look it up in a dictionary, a physical dictionary, you remember the process in which you turn a dictionary, you look for the word A and C and A and R, you remember perhaps the smell of the dictionary and the position where it is on the page.
And then so we remember by association, whereas in comparison, if you look it up in an electronic dictionary, that process is stripped and it happens so fast.
But Lee mentioned one application or one example of one specific technology that these AI devices are using, namely optical character recognition, which allows the device to tell and recognize questions that students upload.
But I'm sure there are other features.
We are talking about very powerful AI education devices that surely they have more functionality than simply telling and finding, pinpointing their questions.
For instance, I believe if it's a successful AI device, they should be able to tell to trial and error what this particular student is good at or weak at.
And then perhaps by some algorithm, they'll be able to arrange a series of tests for them to strengthen their understanding.
Yeah, that's one of the advantages.
I was reading about that as well.
In a traditional classroom setting, the teacher teaches the material, the 30 plus students in the classroom perhaps either grasp the concept or they don't, they get tested on it, and then the teacher moves on to the next level.
But if you're using one of these AI tools and you're struggling with a particular math concept, for example, the AI tool will recognize that and then design some problems for you so that you can strengthen your knowledge of that particular concept.
That's a great advantage.
These devices are locked too.
It's not like kids are going to be playing games all day on these things.
The software is locked, and that means that kids can't download games or other forms of entertainment.
And that's a good thing too, because it means that these are really educational tools.
They're not intended to be anything else, which is a good thing.
But there's something else we didn't talk about yet.
Whether or not the AI tool is really helpful or not is kind of a moot point if the kids don't know how to use it properly, which is kind of a basic thing to talk about.
But that's a very fair point.
There was a 2024 survey in a report here in the country, and they found that only 19 % of young people said that they had taken a specialized internet literacy course.
And even though most of them had gained knowledge about AI tech through online sources or extracurricular classes, they weren't necessarily trained particularly in these or specifically in these devices.
And if we think about economic disparity, where higher income families, those children may have had more access to the internet than perhaps in rural communities.
And statistics show that the device owners are quite low in rural communities.
8 .1 % of minors here in rural China own these devices.
So those kids may not have the same knowledge of tech because they didn't have access to tech before.
Do you see where I'm going with that?
I see a lot of the AI education devices, they run on the Android operating system.
And if you don't have previous experience dealing with an Android device or a tablet, then probably you don't know.
But your example could be an argument for or against the use of AI device because if I would like to use the example to support AI, then I would argue that this is precisely the reason why we want children to be exposed to tech.
Because AI devices, if they are carefully designed, can provide a safer access to the internet as compared to a tablet, which may or may not lead to various or lead you astray, because the type of applications that you can install on the tablet is different from a smart device.
And one of the definitions of AI education device is that many of them, they have a set, a fixed set of applications, only educational applications can be installed on them.
And children do not have the freedom to install, let's say YouTube or TikTok on them.
So the less of a chance for them to be led astray.
Yeah, exactly. Which is you're right, it's a great argument for this.
Again, I hope that these devices are affordable for all families.
I don't think you can put a price ceiling on AI.
Or there should be a price ceiling.
I think there might be enough competition in the market where maybe you don't need a price ceiling.
But it is a wonderful tool.
And again, back to the TED Talk, it seems like educators, generally speaking, are in strong support of these AI tools.
Of course, the teachers need to know very well how to use these devices as well, so that the students can come to them when they have questions about this.
Parents need to be involved.
And if everybody's on board and everybody's on the same page, this could really help level the playing field for kids to have an equal right at education.
Sure. And also one last point.
Since the advent of chat GPT, like large language models and AIs, there has been an increase in the number of companies involved in producing these devices.
Whereas previously, only extracurricular tutorial companies were focused on that.
Now tech companies in China, like the iFlyTech, which is more like a voice recognition service, and now they've expanded much further.
So the competition here has become more intense, but then each company must have their own area of expertise.
Some may be good at developing and designing the kind of algorithm that will streamline the process of learning.
Others might come up and brag about their quality resource and the kind of courses that they are able to provide.
So I think it is indeed quite essential for parents to know and to perhaps to go on YouTube to read the latest evaluation of these devices to find out which one might benefit their children the most.
