This week, the newly-formed central government tabled the Union Budget for the year 2024-25. Everything from agriculture to space was spoken of, however, finance minister Nirmala Sitharaman left the ‘AI-word’ unsaid in her budget speech!
According to IndiaAIMission, the Union government has allocated Rs 551.75 crore to the AI Mission, emphasising its commitment to advancing AI research and applications.
Earlier in March 2024, the Cabinet had approved Rs 10,354 crore for the AI Mission, a provision of 10,000 graphics processing units for use by start-ups and universities.
IndiaAIMission mentioned that this financial boost for the next five years will drive key initiatives, including the IndiaAI Compute Capacity, IndiaAI Innovation Centre (IAIC), IndiaAI Datasets Platform, IndiaAI Application Development Initiative, IndiaAI FutureSkills, IndiaAI Startup Financing, and Safe & Trusted AI.
Exciting to see the Union Budget 2024-25 prioritizing DPI, critical minerals, and job creation. The focus on developing DPI applications in agriculture and other areas lays the data foundation for making India the AI hub of the world. The Critical Mineral Mission will be a game…
— Bhavish Aggarwal (@bhash) July 23, 2024
Is this Enough?
Considering the global AI race, the government must step up its game.
In March 2024, US President Joe Biden’s budget allocated $202 billion for federal research and development (R&D), recognising the potential benefits of AI while addressing its risks.
This budget includes a significant investment of $729 million for AI R&D at the National Science Foundation (NSF), marking a 10% increase from the 2023 budget. From this fun, $30 million is designated for the second year of the National AI Research Resource Pilot.
Further, it includes $310 million for DARPA’s AI Forward initiative to research and develop trustworthy technology that operates reliably, interacts appropriately with people, and meets the most pressing national security and societal needs in an ethical manner.
Further, the UK has announced £100 million in AI funding for the Alan Turing Institute over the next five years, as part of its Spring Budget 2024.
India’s AI Research Game
At the IGIC 2024 event in Bengaluru, a research scholar from Singapore highlighted a curious trend: grants were being awarded to studies that merely mentioned AI. Interestingly, India’s AI research scene has seen a dramatic rise lately.
In 2023, India published 17,000 AI papers, a significant increase from fewer than 4,000 just a decade ago. This surge pushed India past the UK, which previously held the fourth position.
However, the question is how relevant are these papers?
Several researchers aimed to measure the difference in research quality. In 2020, Klinger et al studied the development and variety of topics in AI research and looked at how much private companies influence this research. They analysed around 90,000 AI-related papers from arXiv.org.
Later in 2021, Hagendorff and Meding analysed 10,000 machine learning papers from three major conferences (CVPR, NeurIPS, and ICML) published between 2015 and 2019. They categorised the papers by searching for academic and industry-related words in the full text.
Their findings showed that the number of papers involving both academia and industry is increasing. They measured the impact of papers by their median number of citations and found that industry-affiliated papers received more citations than academic ones.
However, this difference decreases over time as newer papers accumulate citations more slowly.
They also found that academic papers were about two years behind industry papers in mentioning trending topics like “adversarial” and “deep learning”. An analysis of social impact terms showed no significant difference between academic and industry papers.
What’s Next?
India serves as a great playing field to experiment with any new innovation.
For AI to succeed on a large scale, access to diverse data, which is abundantly available in India, is required. The diversity within the population acts as a natural protection against bias, allowing for more accurate and representative AI models.
“If AI works in India, it can work anywhere. This is because the diversity within the population acts as a natural protection against bias, allowing for more accurate and representative AI models,” said Shekhar Sivasubramanian, the CEO of Wadhwani AI.
But sitting on diversity advantage with no money will take the country nowhere.
