UAE recently launched an Arabic language model Jais, which is an open source model containing 13 billion parameters built on a trove of Arabic and English-language data. With this launch, the UAE has elevated its position in the global AI race. In May, Abu Dhabi’s Technology Innovation Institute (TII) launched UAE’s first large-scale open-source model, Falcon 40B, and it ranked 1st in Hugging Face’s Open LLM leaderboard.
A few days ago, China sprung back in the global race of AI chatbots. Five Chinese tech firms — Baidu (Ernie bot), SenseTime Group, along with AI startups Baichun Intelligent Technology, Zhipu AI and MiniMaz launched their AI chatbots to the public. Considering that it is a tightly regulated market where companies are required to submit security assessments for clearance before launching AI products to the public, the release of these models could be interpreted as a strategic move by the Chinese government to compete globally, particularly against the US.
Considering how it is a strict market and companies need to submit security assessments for clearance to launch AI products to the public, the release of the models can be construed as China government’s push to take on global competition especially the US. Interestingly, within 12 hours of its launch, Ernie bot ranked 1st in popularity in Apple’s app store in China. Baidu also plans to release ‘AI-native’ apps.
Amidst such progress, does a country’s ecosystem facilitate this type of development, and where does India stand globally in such a race?
Source: Stanford University AI Index Report
Indian Government Needs to Accelerate AI Investments
While countries are heating up in the global race, India stood at 8th position globally in terms of private investments in AI.
Earlier this year, in the Union Budget 2023, Nirmala Sitharaman announced that three centres of excellence for artificial intelligence will be set up in top educational institutions. Pushing for a vision of ‘Make AI in India’ and ‘Make AI work in India,’ the Finance Minister emphasised on industry players partnering with research for AI solutions. However, there was no budget allocated from the government’s end to facilitate any type of AI development.
Pushing for CoE in educational institutes alone might not lead to immediate development. Research institutes such as AI4Bharat are already doing that, but to have an accelerated pace of development, private players need to be involved. This would also involve setting up AI infrastructure projects, incubators, AI dedicated tech parks, and probably doubling down on AGI research funds.
In the US, there are special exemptions for companies that invest in green and sustainable energy. Thereby inviting a number of big tech leaders investing in the segment. Offering incentives or tax redemptions of any kind to investors that can contribute in the AI space can help. Considering how there are tax exemptions for startups under DPIIT, introducing a niche segment for AI investments can boost growth.
In the recent G20 Summit, Narendra Modi called for a global framework on cryptocurrency, artificial intelligence (AI) and other emerging technologies. Ironically, no government investments on crypto or AI have happened.
Uncertain data regulatory measures in the past had put big tech companies on a cautionary mode, however, with the Digital Bill that was passed recently can help streamline future investments.
The government needs to quickly step up AI investments else it would lead to a missed opportunity. Similar to how we arrived late in capitalising on manufacturing smartphones, semiconductors and GPUs, the same should not be repeated for AI and quantum computing.
While the government needs to step up, efforts are materialising in the private segment.
Atmanirbhar Bharat?
IBM Chairman and CEO Arvind Krishna in his recent visit to India for the B20 summit, said that India should develop its own sovereign capability in AI, and consider establishing a national AI computing center. He even suggested that the government and private companies should be able to leverage computing and data infrastructure in ways that can harness AI for the country’s distinct goals and requirements.
While over-ambitious plans are always in the making, there are obstacles that hinder big tech companies from freely operating or accessing information from the country. In May, Rajeev Chandrashekar, said that the foreign firms like Google have to invest in Indian AI startups in order to access the country’s data set, and that the government’s focus is in supporting intellectual property creation by domestic deep tech startups in AI and semiconductors.
The government has always been clear in promoting development within India. Furthermore, as per a Linkedin report, India has the highest AI skills penetration in the world, which is an indication of the rich talent pool that can be capitalised.
For accelerated growth, an effective collaboration between government and think-tanks or relevant private players might be the way. For instance, the latest Arabic model was launched through a collaborative effort – California-based AI company Cerebras in collaboration with Mohamed Bin Zayed University of Artificial Intelligence, and Abu Dhabi-based technology holding company G42.
Last month, Cerebras Systems had signed a $100million AI supercomputer deal with G42 committing to delivering nine AI supercomputers. Cerebras’ supercomputer Galaxy Condor was used for training Jais.
Such levels of partnerships and support are sprouting in India as well. AI4Bharat, a research lab at IIT Madras, has been on the forefront of building Indic large language models. AI4Bharat is also supported by Microsoft. In collaboration with the big tech, a generative AI-driven chatbot for government assistance called Jugalbandi was launched. The research lab is set to raise $12M funding from Peak XV and Lightspeed Ventures.
Slowly Materialising
Just when we were thinking how Indian tycoons were shying away from generative AI investments, there has been a string of announcements from Indian entrepreneurs and tech leaders in the last couple of weeks.
Tech Mahindra announced Project Indus, an indigenous large language model that would support multiple languages. In the initial phase, the model aims to cover 40 Hindi dialects, and people can contribute to the model by lending their voice via specific prompts.
Mukesh Ambani, recently announced Jio’s plan to develop India-specific AI models that will benefit various sectors including government, business, and consumers. The company also intends to invest in creating 2000MW of AI-ready computing capacity.
While private players are waking up to actively spearhead an AI revolution of sorts, unless the Indian government actively steps in, advancements will only slowly move. With AI estimated to contribute massively to a nation’s economy, India’s AI investment at the right time can push it ahead in the global race – as of now, AI is only talk no action for the Indian government.