Every time I have published a paper in the last six months, I have experienced a phenomenon that has deep implications for the future of higher education. Within about three hours of publication, someone would post a podcast based on my paper. It sometimes involved one person delivering a lecture about the paper, and sometimes two people discussing it. The quality of such podcasts is surprisingly good, including their grasp of technical issues.
The really interesting part is the voices are not human, but of bots generated by artificial intelligence. Yet, they are completely life-like. They emphasise the key points and even have a touch of humour. Indeed, the voices include small imperfections of speech that give it a truly human quality.
The AI algorithms seem smart enough to go beyond a simple regurgitation of my papers. They go on the web to find simple definitions of the technical terms I have not bothered to explain, and even look up information from the papers I have referred to but not written about explicitly in the main text. Thus, the AI bot is capable of intelligently extracting wider information to add value on its own. It’s also capable of conducting a question-and-answer session. In other words, AI is already capable of delivering a lecture about my research that I would myself find hard to match.
It does not end here. AI models are already capable of absorbing material from a paper or a podcast, and convert it into an exam paper that tests human comprehension. It can then mark the test, identify the gaps in understanding and recommend corrective learning. All of this can be done almost instantaneously, at a tiny cost. Note that this technology is not in development—it already exists.
The implications are obvious: AI-based bots will soon be able to replace human lectures in the most standardised subjects. They will also be able to answer questions and conduct exams in many fields with a little bit of human supervision. Human interaction will still be needed in non-standard areas of knowledge, or areas requiring hands-on skills—say, in surgery—but it is clear that many areas of knowledge can be substantially automated.
Students in primary education will probably still need a lot of human hand-holding, though digital material will help here, too. But it’s likely that by class 11 or 12, students will be increasingly relying on AI-generated material. Moreover, by the time a student is in college, she will be gaining most of her knowledge from digital systems. Whether or not academics like this disruption, this shift is already taking place. Rather than fight the inevitable, we should embrace the opportunities it presents.
It is not easy to predict the long-term impact of such non-linear shifts, but here are some likely implications. First, it will flip the use of lectures and hands-on skilling. Today’s classes are mostly lecture-based and projects are usually ‘homework’. In the future, students will be expected to listen to lectures at home. Students already do this in many courses, and come to class to mark attendance. Their class time, therefore, should be used for applying the knowledge to solve problems and learning team-work. This will eventually blur the line between the university and an industry apprenticeship.
Second, with high-quality information and teaching tools freely available online, college education need no longer be about spending a fixed time at the university at a certain point of life. Most college degrees will eventually be about collecting a specified number of credits. A student should be able to do it at any point in life, and at any pace that suits her. The University Grants Commission is already enabling this approach by allowing multiple college intakes in a year while giving students the flexibility to finish their courses at different speeds.
Third, academia will have to re-orient away from lecture delivery to research. Compared to global peers, India’s university system does not emphasise the generation of new knowledge. Since old knowledge is anyway commoditised by AI-based tools, academia will have to focus on the generation of new knowledge and its application in the real world.
Meanwhile, advances in knowledge will also come from many sources outside academia. The gap between academia, industry and even talented hobbyists will blur. This needs a change in mindset—including for research journals that today publish articles based on authors’ credentials, rather than the quality of their ideas. In this context, the Centre’s ‘One Nation, One Subscription’ policy is a great initiative, as it makes the latest knowledge freely available to a large number of researchers.
Fourth, the new tools present a big opportunity for India to make tertiary education cheaply available across the country. Since students in most fields will need to spend only limited time at the university, it will allow the existing infrastructure to be leveraged many-fold. This would partly solve the problem of ramping up a large number of institutions in time for the demographic spike of the next quarter century, followed by over-capacity in the demographic slump beyond. East Asia suffers the problem of having institutions that were too late to ride the regional demographic peaks and are now empty.
Fifth, AI-generated tools could make education language-agnostic. All knowledge will be available eventually in most widely-used languages, since the tools will seamlessly translate the material. A lecture delivered in Swedish would be available in Bengali within minutes. If there is a problem with a technical word, the AI-bot will explain it with infinite patience to an interested listener.
Finally, given the rapid changes in skills needed in a technologically and geo-politically fluid world, we will simply not be able to keep up using the traditional system of teacher training and curriculum upgrade. Only AI-based systems will be able to quickly create curriculums, teaching material and testing systems in fields that do not exist today.
To conclude, AI-based learning systems will soon disrupt the current lecture-heavy system of higher education. This is now unavoidable and the Indian academia should embrace it. It provides the opportunity to make quality teaching and certification cheaply available across the country, in multiple languages. Freed of repetitive lecturing, academia should focus on research and development in collaboration with industry and private talent.
(Views are personal)
Sanjeev Sanyal | Member, Prime Minister’s Economic Advisory Council