The AI Engine

As the placement representative of the first batch of students who had gone through a Master's programme in data science and artificial intelligence (AI) at the Indian Institute of Technology (IIT) Madras, Tanishq Singh was acutely aware of an element of trepidation among a few of his classmates. The programme was established in 2022 in collaboration with the University of Birmingham and adhered to MSc standards in the U.K.; however, students were concerned whether it would be treated on a par with an engineering Master's degree. "The hesitation was especially among people who had not done a BTech in CSE (Computer Science Engineering)," says Singh. "Being the first batch, we were not sure how many companies would offer us a job for this specialisation." However, these apprehensions were laid to rest by the time of the convocation: almost every student had been recruited as an AI engineer, AI consultant, or data scientist in start-ups and software companies.

The MSc course was among a series of new programmes that IIT Madras had started in the past five years, in line with the trend in other IITs as well. The institute set up the Robert Bosch Centre for Data Science and Artificial Intelligence in 2017 to use technology for decision-making in engineering, finance and healthcare. The centre brought visitors and ideas to the institute, along with information on what industry was looking for in the country. In 2018, IIT Madras started programmes that would provide a dual degree to an engineering student with an additional year spent on learning another subject that would enhance their career prospects. The dual degree programme in data science was the most popular.

The curriculum was designed in a way to enable students from any discipline to spend an additional year for a Master's degree in the subjects offered. In the first year, for the dual degree in data science, IIT Madras admitted 25 students. This number went up to 30 the next year, 50 the year after, and 75 in the fourth year. Students came in from all disciplines, a feature that is seen in the PhD theses in AI topics as well. In fact, the best data science PhD theses came from a variety of departments, such as computer science, electrical engineering, and biotechnology. Research in data science and AI was spreading across departments, partly because of the increasing demand from industry for such work.

Indian industry was hungry for trained people in AI, and eagerly embraced those with such expertise. India had seen the ratio of AI talent hiring to overall hiring grow 33.4% in 2024, according to the Stanford University's AI Index, based on data from networking platform LinkedIn. The report also put the number of newly funded AI companies in India at the fourth position globally, behind the U.S., the U.K., and China. A 2025 report from Quess Corp, which deals with staffing and workforce solutions, estimates that only one of 10 vacant GenAI engineer positions in India was filled by the country's workforce. Universities across India, from premier institutes to Tier-2 and Tier-3 colleges, are responding to industry needs, reflected in the rising number of undergraduate degrees and specialised programmes in AI. About 10 IITs today offer a standalone BTech in AI courses; the first one was started at IIT Hyderabad in 2019.

As Indian institutions create a new stream of education from the ground up, academicians and industrialists also hope that the new breed of graduates will do what computer science graduates accomplished several decades ago: invent a new industry.

When scanning the AI education landscape in India, academicians find similarities to the early days of the computer revolution, when computer science branched off from engineering as a separate discipline. There were few computer science graduates in India when the information technology (IT) industry started looking for programmers at the end of the 1970s. The first set of companies hired anyone with a good technical background and trained them to do programming. Soon, training institutes such as NIIT mushroomed and provided programmers with the basic skills needed for the IT industry. The leadership in the early days also came from a variety of backgrounds, but rarely with a computer science degree. Students learned some computer science as part of electrical engineering, as it was not taught as a separate discipline.

A similar shift is underway today in the field of AI. AI tools developed in India are created by those who have moved from related disciplines, and those who deploy the tools are the AI analogues of the NIIT graduates in programming. However, as happened with computer science in the 1980s and 1990s, the IITs and other premium institutions are now creating specialists with the knowledge and skill to develop AI products. Some observers expect them to start a new set of companies, just as computer science graduates did earlier.

When computer science first branched out, the hardware component of computers was a key part of the curriculum; electronic design, semiconductor technology, microprocessors were all taught, as was communication theory — how to transmit signals over radio and so on. It was only in the 2010s that the IITs removed the course on principles of communication from the computer science curriculum. "Even years after universities started teaching CS (computer science), I remember there were diehard advocates arguing for the field to get reabsorbed into electrical engineering," says B. Ravindran, who heads the Wadhwani School of Data Science and Artificial Intelligence, at IIT Madras. (He himself started with a degree in electronics and communication engineering and went on to specialise in computer science.)

Computer science, however, borrowed heavily from other branches too: crucially, it needed aspects of mathematics and logic to think about data structures and algorithms. "As computer science started evolving, abstraction started getting built in. For example, when talking about computer networks, you didn't have to worry about how signals are modulated; you thought more about higher-level protocols like TCP (used to transmit data over the internet). Details of the physical layer receded completely into electrical engineering," Ravindran says.

The new field of AI is far more diverse and eclectic than computer science. It is influenced not just by computer science and mathematics; it also relies on work from other disciplines such as philosophy, neuroscience, economics and logic. While AI was taught as a specialisation under computer science before 2020, it was during the breakthroughs in Large Language Models and the launch of ChatGPT that the All India Council for Technical Education (AICTE) approved standalone BTech programmes in AI and Data Science, prompting many universities to start a special department for the field. This also legitimised online certification courses offered by colleges and the IITs.

The idea of a standalone AI curriculum, however, is still finding its feet, with a certain ambiguity about subjects that should be included in it. A standard curriculum for an undergraduate degree starts out with foundations of machine learning, applied statistics, and computational theory — subjects that will go on to help students build AI models, and not just know how to use them. Deeper specialisation in progressive semesters will focus on the various branches of AI such as reinforcement learning, natural language processing and computer vision. Some institutions have also seen the need to integrate AI with human culture and society.

For IIT Kharagpur, which started its first batch of BTech in AI in 2024, a new department for the discipline became necessary as the field found its way in domains beyond engineering: from medicine and judiciary to agriculture and weather prediction. The institute's leadership was dissatisfied with the surface-level education of AI, as it was not producing a deep reflection into AI and its effects on humanity and its ethics. "The larger subject of what we can call Humanity and AI needs to be answered through the evolution of curriculum and research programmes," says Director Suman Chakraborty. "Covering some of these more sensitive or intricate aspects of AI would not have normally been possible with the standard major-minor elective route." To build the new curriculum, the foundational courses were an easy choice: the institute already had a Centre for AI, and it drew from the centre's experience. However, adding the exploratory courses in the form of electives required more thought, especially because the field was changing rapidly. For example, writing code was not a big differentiator anymore. In computer vision, techniques related to extracting features from images are not sought after so much because of the advancements in deep learning.

Some of these trends had begun in the top global universities a few years earlier. In 2018, Carnegie Mellon University (CMU) became the first university in the U.S. to offer a Bachelor of Science in AI. Gradually, its computer science department was reorganised to include undergraduate programmes in not only computer science and AI, but also computational biology, human-computer interaction, and robotics. According to Martial Hebert, Dean of Computer Science at CMU, AI as a separate field was being discussed in the department much before 2018. "The reason we thought it's important to have a degree specifically for AI," he says, "is that we get to shape the curriculum according to what we believe are skills a student must learn to advance in this field. That is very different from a student taking electives that they choose because they're interested in it." The aim is to educate students to build AI models rather than just deploying them.

The popularity of AI degrees has risen steadily at CMU, with a doubling of AI students from 10% to 20% of the computer science department in five years. A similar trend is seen in other universities in the U.S. It is being witnessed in India as well. The highest rank of a student in JEE (Joint Entrance Examination) who chose an AI undergrad at the IITs was 135 in 2024; in 2025, it was 102. The lowest rank similarly shifted from 414 to 298 over the same period.

Vasudeva Varma, Head of Language Technologies Research Centre at the International Institute of Information Technology, Hyderabad (IIITH), has a prediction for AI education and research for the next few years: it will start integrating into other fields, including humanities and social sciences. This is somewhat like the way computer science moved into all different fields over a period of time, with the difference that AI touches far more fields. "Social aspects of AI should also come into play in this field, and we need greater collaboration of AI researchers with humanities and economics, to better control the role they play in human society," he says.

Academic practice has already started preparing for this by bringing AI to diverse disciplines. At the moment, colleges follow the major-minor or double-major elective system. This means that even if students have a degree in a completely different stream, they can choose to add an AI or data science minor along with it. The AICTE has already included AI modules in the electrical engineering curriculum, starting this year. By 2026-27, it will roll out a revised AI-integrated curriculum for most domains, including mechanical, civil, aerospace, biotechnology, chemical, textile, and agriculture engineering.

Sashikumaar Ganesan, who leads the AI for Research and Engineering eXcellence (AiREX) Lab at the Indian Institute of Science, Bengaluru, finds this horizontal aspect of AI education even more important than studying core AI. "To apply a solution, you need domain expertise," says Ganesan. "For example, machine learning is needed in mechanical engineering, chemical engineering, bioengineering... Just one dedicated department studying ML is not enough," he says. He compares AI to foundational engineering mathematics, learnt commonly across all departments. "This is already being done in a lot of universities across India." At IIT Madras, the AI programme includes domain-oriented courses such as computational chemistry and biology. Here, they would learn to use AI techniques to solve problems in manufacturing, transportation or healthcare.

So far, those who work in AI have learned their job through trial and error, by thinking on their feet. When Spandana Raj Babbula joined Google nearly a decade ago, she moved around departments such as Ads and Voice Assistant that needed her to understand how to handle large-scale data in a cost-efficient manner. But when GPT entered the market, she quickly realised the possibilities for innovation in LLMs and sought to move into that field.

"To do this pivot, I had to pick up many things on the job — even the ability to program accelerators and use frameworks like JAX," says Spandana. Now she sees AI undergraduate students learning these skills as part of their coursework, even before they enter the workforce. "When I was in college, we did not put so much thought into it; we just chose the electives we found interesting," she says. While working, she found that the subjects were all interconnected; one skill was built on top of another. "Half of the burden for the student is taken away because somebody has thought about all of the subjects in a holistic manner and put them together in one degree," she says.

Today, Spandana is part of hiring panels for Google, and observes that most tech companies look for freshers with strong mathematical foundations, an understanding of various classes of models, an ability to program accelerators and also a strong systems foundation. While an AI programme that also equips students with strong programming skills might give them an edge now, it will no longer be a differentiator after the degree becomes mainstream. "The field is evolving so fast that whatever you learn now is never going to be enough."

Institutions are also trying to avoid the mistakes of several decades ago, when computer science was slow to take off as an independent discipline, with consequences for Indian industry. "We were busy developing the services industry and could not create deep tech giants. But having learned that lesson, we now have an opportunity to not make similar mistakes," says Varma of IITH. "Do we have an educational ecosystem to produce top-class AI scientists, engineers and researchers? That will determine whether we have a long-term plan or a short-term plan."

See also:

Course correction
Testing the waters