Incentives for AI in education in China
As AI technology becomes more embedded in education and the nature of education changes, what are the repercussions and points of divergence between China and the West? (7 min read)
It is commonly accepted that both the nature of education and the toolkit that is used in educating is changing at a rapid pace, particularly as the nature of work and economic growth evolves. Technology has come to play an increasing role in education, especially in China. Within technology more broadly, AI (especially machine learning applications) is changing the nature of what it means to educate. The conditions for education in China are extremely supportive of the application of AI.
Despite increasing rhetoric around trends such as the aging population of China, it is important to put the scale of education requirements in perspective.
With almost 260 million students and over 15 million teachers in about 514 000 schools (National Bureau of Statistics of China, 2014), excluding graduate education institutions, China’s education system is not only immense but diverse. Education is state-run, with little involvement of private providers in the school sector, and increasingly decentralised. (OECD Education in China - A Snapshot)
Tools that encourage the use of AI to improve efficiency in K12 education are most commercially successful in China
The AI journey in China seems to be focused on efficiency and optimisation of learning content that lends itself to objectivity (think maths). For example, Squirrel AI, an “adaptable” learning platform that allows students to create a personal journey through a pre-defined syllabus, gradually modulating the questions that they receive to focus on their areas of weakness. Unlike more traditional K12 technology offerings that break down a syllabus into, say 3,000 points, Squirrel AI has an order of magnitude greater number of data points - closer to 30,000. This enables a much greater degree of real time analysis and as a student uses the platform and quickly hones in on the aspects of their knowledge that is deficient.
The main focus to date has been on standardized tests, which China has been obsessed with (notably the 高考 university entrance exams) and that too on tutoring, rather than being a formal part of the education system. Data points on student behaviour are shared across the platform generating a form of aggregation economies and network effects. The unique approach here also reflects the desire of the company to see teachers becoming “pilots”, rather than central actors, so students are essentially taught in a personalised way by their algorithms.
One might expect that students would initially be averse to studying more but it turns out that they love it! Why? Because it actually cuts down the amount of time that they need to dedicate to studying since they are able to achieve increases in test results much more quickly. The only complaint seems to be around the lack of human interaction with a teacher.
So why is AI in K12 education flourishing?
This is a systemic concern for China as a whole, as reflected in the current 5 year plan given the potential to accelerate growth and importantly, bridge the urban-rural divide and therefore achieve a higher equilibrium level of prosperity for the country. I would expect some variant of it in the next one. “Good quality” education is critical to achieving China’s desire to move up the value curve, especially with respect to technology, which is a source of strategic future global competitive advantage. A focus on STEM is expected.
The amount of data available is extraordinary, as well as a generally higher degree of willingness to use it. Additionally, there is widespread adoption of new technology in all aspects of people’s lives and the marginal cost of computing is incredibly low due to a well-established cloud computing infrastructure and tech labour pool.
K12 education is by definition, time-consuming and has been labour intensive too (historically, there has always been a ratio of teachers to students for effective education). AI offers the ability to circumvent this constraint very effectively.
K12 education is critical to every family in China. In the media and amongst the Chinese mainstream, any reference to education is synonymous with anxiety. There is huge societal pressure to succeed through education. However, the standard of schools varies massively from province to province. Good quality tuition costs a lot of money. The cost of housing in many “first-tier” Chinese cities and the competition for good jobs for the average person means that there is truly very little disposable income available. In this situation, and given the historical tendency for Chinese to save and invest, it is natural that the value placed on education should be high and that parents would want to have a quick and high, measurable ROI for any additional outlay on education (which Squirrel AI ostensibly provides given the evidence in the form of better results) at a price point that is viewed as being reasonable. At the same time, even though the purchasing power and disposable income of the rural population or those in “lower tier” cities has increased dramatically, there is still an over-riding cultural sense of elevating one’s position in the world through the power of education, particularly in STEM subjects.
Bearing these points in mind, it is unsurprising that AI has begun to be aggressively introduced into the Chinese educational system.
Given the factors above, it is not a surprise that this approach has gotten the most traction early on (Squirrel AI is amongst the most highly valued of all Chinese ed-tech startups). It essentially enables the Chinese state to use a private sector entity to massively accelerate basic technical learning, a key concern for most families at present, and generate valuable intellectual property and know-how that could be exported (indeed Squirrel AI has ambitions to expand to the US and Europe). This approach, taking into account the coherent and longer-term, industrial policies of the entire nation is a key differentiator with respect to China and the West, which does not have the same intensity of coordinated goals (even though there are signs that this may be changing).
These incentives point to a much more and ongoing rapid adoption of AI in China than may be expected.
What hasn’t happened yet?
AI has not yet been applied to educating people for what is becoming seen as the future of human competencies where we have increasing technology augmentation - creativity, empathy, collaboration and so forth. This is where “responsive” AI can come into play, essentially creating a bespoke environment for each student to learn these key skills, together with human facilitators and guidance. China does not seem to have focused on this element of education yet (and neither has the West to any great degree), partly because of the sheer number of people that need to be educated to a level, the automation of white-collar roles has not picked up pace and this is clearly a more difficult problem to tackle (i.e. there are other low hanging fruits out there). However, as soon as the incentives for this become apparent, which I believe will happen within the next 2-3 years as the returns from “adaptive” AI become realised, expect this to become even more rapid.
What are the main implications to be aware of?
Expect much more in the way of AI to be applied, much more quickly than may be expected. Every part of every subject that can be atomised into its constituent parts will be done so, tracked and analysed. We clearly do not know the margin structure associated with these types of AI companies in China but I would expect them to follow the model of most SaaS companies with some form of network benefits (in this case, the increasing marginal value of data) - high incremental margins. Chinese companies will have a sustainable growth and product development advantage over the rest of the world.
Education in China will be democratized, which also has ramifications for where people chose to work and live.
The technologies and mechanisms developed in China will be aggressively exported to the rest of the world, especially other Emerging Markets. If governments around the world can buy Chinese surveillance technology and 5G infrastructure (even the UK!), why not Chinese education tools?
Given the politically sensitive nature of education, I would expect market entry by Chinese AI companies into markets like the US to follow partnerships with established education technology companies.
There will be no shortage of capital willing to fund these types of businesses. Expect large amounts of venture capital to continue to flow into this space, as well as public listings by these types of companies.
Do you like what you read?