Funding remains one of the most significant challenges in accelerating AI research in India. Compounding this challenge is a lack of motivation among researchers to pursue revolutionary innovations. However, university researchers alone can’t be held responsible for this, especially when compared to the West – where strong ecosystems enable the development of great solutions.
To give an example, professor and university distinguished scholar of CS and engineering at Ohio State University, Dhabaleswar K (DK) Panda, presented his newly-established $20 million (National Science Foundation) NSF-funded AI Institute Intelligent Cyberinfrastructure with Computational Learning in the Environment (ICICLE) at the IEEE HiPC 2024 event in Bengaluru this week.
Panda, who is also the founder of X-ScaleSolutions, highlighted how his initiative is focused on building the next generation of high-performance compute for big data, machine learning, and the future of deep learning. The focus is simply on the democratisation of AI solutions instead of keeping it in check by a few big players like Microsoft, Google, and others.
Panda is focused on solving many use-case problems, such as those in agriculture and manufacturing. To scale this solution beyond the US, Panda is looking to partner with several Indian institutes, such as IIT Bhubaneswar and TCS Research.
Against the backdrop of the HiPC event, AIM spoke with Panda to better understand the project and asked why such initiatives are rarely seen in the Indian research ecosystem. Panda agreed with much of this, saying that things are slowly changing for Indian universities.
Too Much Focus on Research Alone
The biggest factor is the massive funding gap between Indian and Western universities. Research in India is often constrained by limited resources, which makes it hard to take on long-term or foundational projects, especially in fields like AI.
The ICICLE project focuses on researching open-source projects that can motivate researchers from across the globe to solve problems. This is similar to something AI4Bharat or other such initiatives are building in India, but this has only been the case for the last few years.
“The ideas are there [in India]; it depends on resources, manpower,” Panda said. He highlighted that since ICICLE got this funding from the National Science Foundation, it was able to put together the right team with the right expertise.
“Thirty years back, the gap between US universities and Indian universities was huge, but now that gap is shrinking. Most of it is about funding. As long as the university is well funded, it will scale since the talent is here.”
Panda, however, added that the researchers need to be trained so that they can move to the right field at the right time. “US universities are still focusing on fundamental research…If a student or faculty gets funding, they focus on short-term development.” Expressing optimism, Panda said he believes the most interesting research in AI could soon come from an Indian university.
A Common Sentiment
Agreeing with Panda’s sentiment, Amit Sheth, the chair and founding director of the (AI Institute, University of South Carolina) AIISC, told AIM that few institutions are increasingly putting students’ research first, such as IIT-D, IIIT-D, IIIT-H, IIT-M, IIT-P and IIT-Mumbai. Only a handful of universities are able to publish research in top conferences.
“In the USA, all the projects they get to work on involve advancing the state-of-the-art (research),” Sheth said. He also highlighted the issue of a publication racket prevalent in India and several other developing countries, with only a handful of researchers from select universities standing out as exceptions.
Commenting on the disparity in research quality, Sheth noted that the gap is non-existent or very small in the universities mentioned above.
“But if you go to your run-of-the-mill NIT, IIITs, technical colleges, etc., the gap is not shrinking and is huge,” Sheth said. According to him, there are some enterprising students at those universities who, in their 3rd and 4th year of pursuing a bachelor’s degree, reach out to US groups to do online internships.
“This allows them to work on the latest topics and often co-author a paper or two, which then significantly improves their chances of getting accepted at top research programs and professors during their graduate studies.”
Moreover, Renjith Prasad, a teaching assistant at the AIISC, told AIM that the core reason lies in the culture. “In India, the focus is so heavily on placements that research often feels like an afterthought. A lot of students I saw preparing to pursue Masters or PhD through Gate did not do it because they’re passionate about research, but because they didn’t land a good job right after undergrad.”
Prasad added that this results in a system where professors are often hesitant to invest deeply in fostering long-term research. “I felt this personally during my Master’s thesis, which was mostly guided by a senior PhD student at AIISC, instead of my own professors in India, and I don’t think I would have completed it without her.”
‘Research Feel like a Solo Journey in India’
Giving his personal example, Prasad said that when he joined IIT Jammu, the first thing he was told was to start preparing for placements through platforms like LeetCode. “The biggest club on campus wasn’t a research or innovation group – it was the placement club.” “Even professors understand that most students are here to chase the best packages, and the few who are genuinely interested in research either head abroad or try to collaborate with someone there.”
AIM earlier wrote that Indian researchers should move beyond PhDs. We pointed out that Indian colleges are generally lagging in terms of where the state-of-the-art is and what they are doing because some of the professors still might be researching in CNN, whereas the SOTA is way ahead.
Pratik Desai, founder of KissanAI, had shared similar thoughts earlier when he said that this requires a fundamental shift from coaching and academia to a change in mindset from parents and founders to investors.
Ritvik G, a graduate student at AIISC, told AIM that the education system was set up in a fashion to churn out developers and self-minded individuals rather than adapting to new times. “From the very moment we joined the university, we were taught that grades mattered more than learning.”
Ritvik further said that the biggest factors for meaningful research in his university in India were an outdated learning environment, limited exposure to AI innovations, limited collaborative support, and a monotonous research environment.
This is often the case with Indian Institutes. “There is heavy focus and overall improvement on the core AI frontier such as NLP and Computer Vision, where most of the recent top conference and journal publications from India exist, but the very lack of diversity amongst teams leads to the development of systems that often are very targeted/outdated in a few years,” Ritvik added.
The post ‘AI Research at Indian Universities Feels like a Solo Journey’ appeared first on Analytics India Magazine.
