Suddenly, everyone is an AI expert. Since the release of ChatGPT in 2022, the field of education has been overcrowded with self-proclaimed AI experts. Every day brings announcements of conferences, training sessions, and events featuring AI experts, along with numerous articles and books about the use of AI in education.
We have so many AI experts, yet very little expertise. Most AI talks and publications reiterate how teachers can use AI in their classrooms and how students can learn with AI. However, very few talk about why education has not improved despite numerous reforms over the past few decades and what AI can truly do to improve education. If AI is used in traditional classrooms to teach the same subjects and meet the same expectations, it won’t transform education. In fact, it’s more likely to become a nuisance than a benefit for teachers and students, creating additional challenges like academic integrity concerns. Under these conditions, AI is unlikely to bring significant improvement to teaching and learning.
Unrealized Potential
Many AI experts and AI systems have advocated for and promised to deliver personalized learning through AI. In practice, “personalized” is a misnomer. Students are allowed to use different methods to show their learning, work in different locations, and go at different paces, but they are expected to learn the same content and reach the same outcomes. This type of learning has nothing to do with the student as a person—a person with individual talents, personalities, experiences, and passions.
Many AI experts, whether from the field of education or technology, seem to assume that schools will continue to operate as they have been operating for the past 200 years. They do not believe the fundamental grammar of schooling needs to or is able to change (Tyack & Tobin, 1994; Zhao, 2024). For instance, there has been very little conversation about loosening the prescribed curriculum to give students more autonomy in deciding what to learn (Zhao & Zhong, 2024). Instead of reimagining school structures, our AI experts often simply layer AI onto the existing system and expect transformative results. The same thing has happened with previous advancements in technology. As Seymour Papert, founding faculty member of the MIT Media Lab, argued decades ago, unless schools change, the power of technology cannot be realized (Papert, 1999).
Papert is correct. Today, despite the massive investment in technology in schools and significant changes since the arrival of computers and the internet, schools have not changed enough, and thus technology has not delivered on its promise to transform learning.
If AI is used in traditional classrooms to teach the same subjects and meet the same expectations, it won’t transform education.
Similarly, the current “copy+paste” way of thinking about AI in education actually undermines AI’s potential. Traditional teaching can of course be improved by AI, but the return is tiny because teachers have been trained to teach traditional classes well without the use of technology. AI can do a lot—from serving as a learning partner with students to supporting administrative work to reshaping teaching—but to truly have an impact in education, it demands thinking beyond incremental improvements in schools: it demands a transformation.
There is a saying, attributed to Seymour Papert, that if a jet engine is connected to a horse wagon, you can either follow the horse wagon by never truly firing the engine or break the wagon by firing the engine. If the potential of technology is to be truly realized, the traditional grammar of schooling has to be broken.
Redefining Personalized and Project-Based Learning
Today, if we are brave enough to realize the full potential of AI tools, which are constantly and rapidly evolving, what changes are needed?
First, we must redefine personalized learning. Personalized learning should not just be about allowing students to have control over the pace or place of learning. Instead, it should be about encouraging and enabling students to discover and develop their own individual strengths and interests. Children are born with different innate capabilities, live different experiences, and are exposed to different opportunities—and they come to school with unique skills, personalities, and interests (Zhao, 2019; Zhao et al., 2022). Traditional schools’ homogeneous curriculum, instruction, and assessment often make students sacrifice their strengths and interests for the sake of meeting standardized requirements. With the help of AI, we could develop a new education paradigm that makes education truly personalized—one where personalization is done by, instead of to, students.
Second, we must redefine project-based learning. There are two major problems with PBL. The first is that the project or problem is often predetermined by teachers or other adults. We expect students to be engaged regardless of whether they have any interest in the work. The second problem is a general lack of interest in the final solution. Too often, the focus of PBL is on the “process” of learning rather than just the final “product.” But without thinking about the quality of the final product, it is difficult for students to go through a process of seeking feedback and refinement, which is necessary for developing meaningful creativity and entrepreneurial thinking. In the age of AI, we should redefine PBL as students finding and solving problems with the assistance of AI tools. Students need to learn how to find and refine problems worth solving and work through a process of innovation to develop solutions, which can be products of any sort, such as music, dances, tools, novels, paintings, and physical objects.
Third, we must redefine schooling. To implement these new definitions of personalized and project-based learning, schools need to be transformed. The traditional age- or grade-based curriculum and standards need to be replaced. When schools move to this new version of personalized learning, they can no longer confine students to the old, prescribed curriculum. Instead, they must imagine a new curriculum, which would not be wholly guided by subjects but rather by student-led personalized learning, problem finding, and problem solving.
With the help of AI, we could develop a new education paradigm that makes education truly personalized.
We in education have not fully explored what a new curriculum like this would look like, but we can imagine some possibilities. For example, this new curriculum could have three parts (Zhao, 2021). One teaches the common or public good, including the skills and knowledge deemed necessary for living in a nation (e.g., government and citizenship). One teaches what local governments, communities, and schools deem important for all students to learn (e.g., basic literacy, numeracy, digital literacy, and financial literacy). The final one is personalized by students.
Redefining Pedagogy, Assessment, and Grouping
For AI to transform education, pedagogy needs to change as well. Traditional class-based group instruction is not necessary to support personalized learning and problem finding and solving, nor is it possible. AI tools and online learning resources are widely available to support student learning, problem finding, and problem solving. Students can seek mentors online from outside the school, as well as work with peers for support and tutoring. With this model, teachers change from instructors to coaches and mentors. They provide students with social and emotional support as well as valuable guidance and feedback (Zhao, 2018a; 2022).
Along with pedagogy, student assessment must be changed significantly. We can no longer subject students to standardized testing and use test scores in any high-stakes judgement about students, teachers, and schools. Since AI tools have already been shown to outperform human beings in virtually all human-made tests, including creativity tests, it is meaningless to force students to take these tests. What schools and school systems should do is develop personalized assessments based on students’ individual strengths and interests as well as their development in problem identification and solution creation. AI can be of great help in reimagining assessments, but systems and schools need to be willing to replace traditional assessments.
Finally, as noted earlier, how students are organized in schools must be transformed. Schools typically place students of similar age into the same classes and grades, which makes it very difficult for teachers to teach effectively because of the diversity of learning needs. Schools have tried to reduce the diversity by taking some students out and placing them in special education, gifted and talented programs, and advanced courses. Teachers have tried to use differentiation strategies to teach to all. But in the age of AI, students no longer need to be organized this way. They can be organized by problems, personalized learning foci, and/or social interests. Or they can be organized differently at different times based on their needs.
Making Big Changes Happen
We have been tinkering with schooling for more than 100 years (Tyack & Cuban, 1995), yet today we are still struggling with improving student achievement in reading and math (National Assessment of Educational Progress, 2021). In the age of AI, we don’t need experts who are simply interested in layering AI onto existing classrooms. Instead, education needs more expertise in understanding how big changes happen so that AI can realize its tremendous potential.
Education needs more expertise in understanding how big changes happen so that AI can realize its tremendous potential.
One possibility to reach this potential, for example, is the “school within a school” approach that starts with inviting willing students and teachers to join a new school within their traditional school. The new school would offer an education completely different from the traditional school but would be operated by the traditional school. This way, the new structures and methodologies can start easily without resistance and can spread because they are exposed to other students and teachers. I have had the experience of working successfully with a number of schools in China, Australia, and the United States using this approach to implement big changes in education.
If we don’t try new approaches like this, we will miss the opportunity to equip our students with experiences of co-evolution with AI tools so that they can thrive with the necessary AI literacy. Moreover, students will not be able to develop their unique greatness—the personal combination of abilities, skills, passions, and creativity—necessary for the age of AI (Zhao, 2018b). So, let’s try to imagine and reimagine the possibilities of transformation with big ideas—big ideas that require big changes.
Teaching Students to Become Self-Determined Learners
Practical strategies to foster autonomy, choice, and motivation in students, empowering them to become self-determined learners equipped to thrive in a rapidly changing world.
