Could AI-centric education be the breakthrough India needs to bridge its learning gaps and equip its massive student population for the future? A Harvard study suggests that students prefer AI tutors over human teachers and that students using AI learn twice as much as those relying solely on lectures.
Researcher and lecturer Gregory Kestin pointed out two trends: advanced students often get bored as lectures move too slowly for them, and struggling students get overwhelmed because they can’t keep up with the pace.
Thus, the time spent on lectures did not correlate with the performance, suggesting that personalised pacing was key to success. The findings also suggest that AI can handle the introductory material, allowing the class time to focus on critical thinking, problem-solving, and collaboration.
This shift could make education more meaningful and applicable in the real world.
Human in the Loop
Despite AI’s widespread adoption, education systems have not significantly changed their teaching or assessment methods.
According to Ethan Mollick, co-director of generative AI Labs at Wharton, AI is not the main cause of cheating. Students cheat because schoolwork is hard and the stakes are high. He argues that the rise of the internet had already diminished homework effectiveness before generative AI.
He believes AI should function as a co-intelligence tool, supporting both teachers and students in thinking critically.
In his blog post, he outlines that while three-quarters of teachers already use AI in their work, most have not been trained in best practices. Research shows that educators who utilise AI for both inputs (understanding concepts) and output (creating quizzes, worksheets, or lesson plans) benefit the most, as it allows them to enhance their teaching rather than simply automating tasks.
Homework also must evolve into AI-powered critical thinking exercises, he suggests. Simulator prompts and interactive AI-driven exercises can encourage deeper understanding by requiring students to apply their knowledge rather than simply regurgitate AI-generated responses.
In his TedX video, Khan Academy CEO Salman Khan explains how AI can help expand education.
Khan revisited The 2 Sigma Problem, written by Benjamin Bloom in 1984. It shows that 1-on-1 tutoring can boost student performance by two levels. This means an average student could become an exceptional one with personalised help. The main challenge has been making this widely available, something that AI now solves by providing personalised tutoring options for everyone.
Why India Could Benefit the Most
The implications of this study extend far beyond Harvard. AI tutors have the potential to democratise education. “The biggest beneficiary of AI-integrated education will be India over the next several decades,” said Surya Kanegaonkar on X.
“The country’s median age in 2050 will be 38 when China’s is 57 and the US’ is 42. TFR will likely level off at ~1.7 in India, meaning that it will have the world’s largest population of students from primary to tertiary education,” he added.
The Case for Localised Solutions in India
The success of AI-driven education is also deeply intertwined with how we understand, manage, and govern data, argues Osama Manzar, founder of the Digital Empowerment Foundation (DEF).
“The first step toward responsible AI deployment in education is mass-scale education on data literacy, information literacy, and digital literacy,” he said. “Citizens must understand how their information is converted into data, how it is used, and the implications of data-driven technologies on their lives.”
Speaking with AIM, Manzar said that civil society must push for the development of localised AI solutions that are designed and governed by the communities they serve. This would ensure that AI-powered education is not just a tool for data extraction but a means of empowerment that respects local contexts, cultures, and rights.
“India’s participation in the global AI movement has largely been reduced to a data supplier and a gig workforce for data entry and datafication,” he said, noting that while this creates jobs, it also sustains a pattern where India functions as a labour force for the world rather than harnessing its own data for national development.
He also stressed that since data could be biased and lead to hallucinations, the importance of humans in the loop cannot be understated. “While AI has the potential to revolutionise fields like education, its outputs are often riddled with biases—linguistic, racial, regional, and cultural—that reflect the limitations of the data and algorithms used.”
