The “iPhone moment” of AI education may still need to wait a while. The conflict between ideals and reality may be the root cause of AI education’s inability to obtain positive feedback. After forming multiple AI teams last year, Chen Xiangdong, the founder of online education giant Gaotu, frankly admitted that they may not have done well, which has raised doubts about the future of AI education among the outside world.
The changing market environment has affected the offline tutoring and online education tracks. Oriental Select has successfully transformed into e-commerce, and Gaotu has also tried knowledge-based live streaming e-commerce. However, “Gaotu Goodies” announced its suspension before this year’s Mid-Autumn Festival holiday. Since then, AI education has become almost the only option for Gaotu to seek a second growth curve. Gaotu is cautious in the field of AI education. This year, it launched the AI tarot psychological healing product “Predicting Tata” and the English learning app for primary and secondary school students “Gaotu AI English.” However, “Gaotu AI English” is behind as soon as it is launched because there are already many similar products on the market, and its emphasized features of “by textbook, intelligent practice, and accurate learning” are not advantageous. In fact, in the past two years, the question in the education industry regarding AI has been whether it can replace teachers. Online education vendors hope to get an affirmative answer to achieve “industrialization” and improve efficiency. However, at this stage, AI’s capabilities are limited and it cannot teach alone.
Luo Bin, co-founder of Gaotu, believes that the relationship between AI and real teachers is a gradual process and can be divided into three stages: Copilot (assistant), Pilot (leading, guiding), and Auto Pilot (automatic guiding). At present, it is almost certain that real teachers and AI combine to serve students, and AI is only an assistant rather than a replacement. But unfortunately, even the auxiliary role of AI in the education field is difficult to achieve at present. The current large AI models have higher requirements for users. It may be an excellent advisor but not a decisive decision-maker. Users need to choose from a vast number of conclusions. If their capabilities are not enough, they may be led astray. When evaluating teachers, the weight of the link of transmitting information in class is not high. The high-value part is attribution, and currently AI cannot achieve accurate attribution, causing teachers to feel at a loss about AI.
AI education is not a false proposition. Under the current circumstances, the best scenario for AI to empower education is the much-criticized “cramming education,” which has advantages in rote memorization and doing practice questions. However, online education vendors like Gaotu are engaged in quality education, demanding both teaching quality and user experience. They cannot promote that AI helps with rote memorization of knowledge points, making AI education products more form than substance and being regarded as “showy” by users. In short, the conflict between ideals and reality is the root cause of online education vendors like Gaotu not being able to obtain positive feedback after betting on AI.
