An innovation ecosystem builder
India’s State of AI Research
India’s Performance in
Top Conferences
Executive Summary
We have analyzed the paper contribution of countries in top 10 AI conferences. Our findings are as follows:
1. India stands 14th and its global share is 1.4% (2018-2023). The US and China lead the world with 21-16x contribution to India
2. India’s AI paper growth rate (CAGR) over 10 years 2014-2023 is 15.5% while Asian economic tiger countries range between 20-30%
3. Alarmingly, India’s growth rate has already flattened, while Asian economic tiger countries have maintained or
accelerated growth further
4. India has maintained or slightly decreased its global share- while it has grown, the world has grown faster
5. 8 Institutes account for 70% of India’s AI Contribution in top conferences. IISC, 5 old IITSs and 2 IIITs are leaders, while new IITs, DTU and ISI Kolkata are rising stars
6. India does better in more applied AI conferences, while it’s share is lower in conferences that accept
theoretical work
Download Full Report
Recommendations
India needs to build capability in basic research in AI other than application development and diffusion. This will help us become a long term leader
-
India should aspire to get at least 5% of global contribution in the next 3-5 years
-
Given that 90% of India’s AI papers come from 20 institutions, they should be the focus to exponentially increase capacity
-
Aim at increasing output at top 20 by 5x in the next 5 years by hiring faculty and increasing PhD students
-
Make sure that compute availability is not a gating factor for doing high quality AI research. Government’s effort in creating computing capacity of 10,000 GPUs is a welcome step
-
Spawn Joint PhD programs at the top 20 Indian Institutes with Universities in USA, Europe and Asian economic tigers for greater collaboration. IIT Kanpur runs one such program with NYU
-
Launch Industrial PhD programs where performing AI scientists in the industry collaborate with Institutions
-
India needs large scale private philanthropic efforts in AI research such as Allen Institute of AI, Machine Intelligence Research Institute, Kyutai.
-
The focus of research must not be only current tech like LLMs and their applications , but also focus on newer disruptions i.e. smaller, interpretable, reliable and controllable models
Parth Mahajan
Jitesh Bhasin
Varun Aggarwal