A controversy over whether India should another large language model (LLM), with two diametrically opposite views aired by high profile personalities, one of whom asked a pointed question, but escaped harsh reactions from people. While one part of the reasons proffered is alternative uses for the kind of monies that will be needed (typically in tens of billions of dollars), it is also a question of whether we have the requisite mathematical skills. Tech Mahindra has claimed having built a LLM in under $5 million but the language used doesn’t suggest a LLM
A muted controversy, when it could have reached very high decibel levels a couple of weeks ago, considering what was said and who said it, with one high profile global personality in the centre. Not just muted but it seems to have died down.
No LLMs please
Recently Nandan Nilekani, one of the co-founders of Infosys and its CEO after Narayana Murthy stepped down, aired his view that India should not pursue building another LLM. This was in response to Google Research India director Manish Gupta’s view that India will benefit from building a foundation model. He reasoned that “If India has $50 billion to spend, it should use that to build compute, infrastructure, and AI cloud. These are the raw materials and engines of this game”.
Some of you may recall an announcement by a leading Indian IT services company in the business media around the time chatGPT had achieved considerable success that they would build a LLM. Nothing has been heard since.
In a sense, this is similar to building a software product but vastly more formidable in terms of investment, manhours and skills. It is generally believed that the kind and level of mathematical skills without which LLMs cannot be build is just not available in India, at least not in the required numbers. In terms of comparison, China has built home-grown giant-sized ecommerce platforms and much more such as civil aircrafts. LLMs need vast investments and data coupled with infrastructure.
Last month, Gupta had said at the Bengaluru Tech Summit that he “respectfully disagreed” with Nilekani criticising that “He is not preaching what he practised”, referring to Aadhar, but completely missing the point in an inappropriate comparison. A not surprising comment came from Rajiv Gangal, VP, VOIS (Vodafone Intelligent Solutions) in a response to a LinkedIn post: He’s always services minded and not aspirational.
Narayana Murthy dismissed (building) because “we have not been able to build large databases, and without big data, AI has no value. Basically, the Indian mindset is still not oriented towards problem definition and problem-solving,” according to a report by Moneycontrol. The Government of India too weighed in with Ministry of Electronics and Information Technology Secretary, S Krishnan, arguing for using existing LLMs to serve specific sectors, but Murthy went a step further: “Which area has (India) invented? Please give me an example”.
Altman’s view
In June 2023, Sam Altman, OpenAI CEO remarked that Indians cannot build something like ChatGPT, because it would be “totally hopeless.”
In a keynote speech at the fifth annual MachineCon GCC Summit in Bangalore recently, CP Gurnani, the former CEO of Tech Mahindra and the founder of AIonOS, claimed they have developed one within five months, that can communicate in about 40 different local languages and dialects, to start. It is called Project Indus. “Sam Altman challenged everybody that India will never be able to have an LLM. I am happy to share that they spent less than five million dollars on what Sam Altman said India will never be able to deliver,” Gurnani said. (https://www.newsweek.com/sam-altman-india-project-indus-1919694)
Nevertheless the language in which this LLM has been articulated raises some doubts whether this is another LLM. Nikhil Malhotra, Global Head – Makers Lab, Tech Mahindra, said, “Through Makers Lab, our R&D arm, we created a roadmap, collected data from the Hindi-speaking population, and built the Indus model. Our collaboration with Dell Technologies & Intel will help deliver cutting-edge AI solutions that enable enterprises to scale at speed. We will redefine the GenAI landscape, driving innovation and operational excellence.”
The emphasis on creating “multiple tailored use cases and enable customers to leverage various applications, including customer support, experience, and content creation across healthcare, rural education, banking and finance, agriculture, and telecom, among other industries”, indicates an application-oriented approach, not foundational. (https://www.crn.in/news/tech-mahindra-launches-project-indus-large-language-model-llm/)
The comments from their partner, Dell Technologies, is suggestive of a system integration approach. Denise Millard, Chief Partner Officer, Dell Technologies, said, “Accessibility and scalability are increasingly important for organisations looking to unlock the power of GenAI. With the Dell AI Factory, LLMs like Project Indus leverage AI-optimised technologies with an open ecosystem of partners, validated and integrated solutions, services and best practices, accelerating the adoption of AI to drive growth, optimise productivity and promote innovation.”
The other partner in the venture, Intel’s comments are also not signalling a foundational approach, emphasising operational efficiency. It is important to recall that Intel has largely left behind in the AI race. In fact, they did not acquire any significant presence in the mobile space too, their success largely confined to PCs.
As recently as October 2024, Tech Mahindra reached a partnership agreement with Zoho and have together struck partnerships with Nvidia to enhance their software offerings through advanced artificial intelligence capabilities. The focus again is on tailoring LLMs for business applications, leveraging Nvidia’s AI accelerated computing platform. (https://www.businessworld.in/article/nvidia-joins-forces-with-zoho-tech-mahindra-to-drive-llm-development-for-saas-and-launch-ai-coe-537158)
Summing up
Surprisingly, there has been no response from the Indian IT industry on tech Mahindra’s claims on LLMs. At least not in public. While every new genuine breakthrough should be welcomed, we must always be wary of marketing some offering as breakthrough. That is sensible intelligence.
