Andrej Karpathy, co-founder of OpenAI, envisions a world where AI robots can become “enthusiastic, skilled at teaching, infinitely patient, and proficient in all languages of the world” subject experts. Through this vision, robots could provide “personal tutoring on demand for all 8 billion of us.”This idea is being realized in his company, Eureka Labs, which is a prominent example of tech entrepreneurs attempting to leverage AI to fundamentally change education.Karpathy believes that AI can address a long-standing challenge: the current lack of talent in schools that are both subject experts and excellent teachers.
In addition to Karpathy, Silicon Valley figures such as OpenAI CEO Sam Altman, Khan Academy CEO Sal Khan, venture capitalist Marc Andreessen, and UC Berkeley computer scientist Stuart Russell also dream of making robots on-demand tutors, guiding advisors, and possibly even replacing human teachers.However, a researcher focused on AI and other new writing technologies has seen many high-tech “solutions” to teaching problems fail. While AI can certainly enhance various aspects of education, history shows that robots may not become effective substitutes for humans. This is because students have long shown resistance to machines, not just because machines are relatively complex, but because people naturally tend to connect with and be inspired by their human peers.
The Cost Challenge of Teaching Writing to the Masses
A director of the English composition program at the University of Pittsburgh oversees the teaching of about 7,000 students each year. Like similar programs he oversees, it has long struggled with how to effectively teach writing to many people at once.The best answer so far has been to keep class sizes to no more than 15 students. Research shows that students learn to write better in small classes because they are more engaged.However, small class sizes require more teachers, which can become expensive for school districts and universities.
Reviving the Scholars of the Past
Enter AI platforms. Karpathy suggests imagining that the great theoretical physicist Richard Feynman, who passed away over 35 years ago, could be revived as a robot to tutor students.For Karpathy, the ideal learning experience is “learning physics materials with Feynman guiding you at every step.” Feynman was known for his approachable demonstrations of theoretical physics, and he could work with an unlimited number of students at once.In this vision, human teachers still design course materials, but they are supported by AI teaching assistants. Karpathy wrote that this team of AI teachers “could run an entire course on a universal platform.” “If we succeed, anyone can easily learn anything,” whether many people are learning one subject or one person is learning many subjects.
Other Personalized Learning Efforts Have Fallen Short
However, personalized learning technologies are not new. Exactly 100 years ago, at the 1924 meeting of the American Psychological Association, inventor Sidney Pressey introduced an “automatic teacher” made from typewriter parts that could pose multiple-choice questions.In the 1950s, psychologist B.F. Skinner designed “teaching machines.” If a student answered a question correctly, the machine would continue to ask the next question. If not, the student would continue to work on that step until they solved it.In both cases, students received positive feedback for correct answers. This gave them confidence and skills in the subject. The problem was that students did not learn much — they also found these non-human methods quite boring, as education writer Audrey Watters noted in “Teaching Machines.”Recently, the education sector witnessed the rise and fall of “massive open online courses” (MOOCs). These courses offered videos and quizzes and were praised by The New York Times and other media for their promise of democratizing education. However, students again lost interest and dropped out of these platforms.
Now, AI-driven platforms are surging. Sal Khan’s Khan Academy launched the Khanmigo AI tutor, designed for “personalized and customized tutoring that adapts to individual needs while accompanying learners as they study.”Education publisher Pearson is also integrating AI into its educational materials. Over 1,000 universities will adopt these materials in the fall of 2024.It can be said that AI in education is not just on the horizon; it is already here. The question is, how effective is it?
Disadvantages of AI in Learning
Some tech pioneers believe robots can customize teaching and replace human teachers and tutors, but they may face the same issues as early attempts: students may not like it.There are also significant reasons for this. Students are less likely to be inspired and excited like they would be by a live instructor. Struggling students often seek help from trusted adults like teachers and coaches. Would they do the same with robots? If they did, what would robots do? We still do not know.Lack of data privacy and security could also become a hindrance. These platforms collect vast amounts of information about students and their academic performance, which could be misused or sold. Legislation may attempt to prevent this, but some popular platforms may not be fully regulated.Finally, even if AI tutors and teachers become popular, there will still be other concerns. For example, if robots teach millions of students simultaneously, we may lose diversity of thought. When everyone receives the same instruction, especially if “academic success” relies on repeating what AI instructors say, where does originality come from?The idea of having an AI tutor in everyone’s pocket sounds exciting. We would also love to learn physics from Richard Feynman, writing from Maya Angelou, or astronomy from Carl Sagan. But history reminds us to be cautious and to closely monitor whether students are truly learning. The promise of personalized learning has yet to guarantee positive outcomes.Source: EDU Guide. The Reading Club is committed to advocating for a reading culture in society and promoting reading for all.