New Delhi, June 22 (IANS) India must build its own foundational artificial intelligence (AI) models to avoid falling behind the United States and China, a new report has said.
The report from global brokerage Bernstein urged policymakers to prioritise domestic model AI development, citing geopolitical restrictions seen on frontier AI systems.
Anthropic’s frontier models were recently restricted for non-US citizens which proved that access to advanced AI cannot be taken for granted.
The brokerage warned that continued reliance on foreign models could leave Indian firms forced to operate with systems “one or two generations” behind those available in the US and China.
Even Indian companies with deep engineering talent could struggle to compete if they have to build products using older AI models while startups elsewhere have access to cutting-edge systems.
“AI is the next fighter jet,” the firm said, adding that cutting-edge large language models (LLMs) are being seen as strategic assets rather than commercially available software.
AI is on the same historical path as technologies such as nuclear energy, defence systems and semiconductors took, where access to such technologies were shaped by geopolitics rather than free markets.
If India continues to depend on foreign AI models while focusing only on application development, it risks ceding control over a critical layer of future technology, Bernstein said, adding that the country’s AI future cannot be built on “borrowed models”.
It attributed India’s historical technology ecosystem being driven by IT services rather than consumer internet platforms as the reason for falling behind in development of foundational platforms. Foundational models require large proprietary datasets to train frontier models, and lack of such data sets led to policymakers and industry leaders arguing that India should focus on AI applications.
India can develop specialised domain-specific LLMs using proprietary datasets in sectors such as healthcare, industrials and defence, the brokerage said.
—IANS
aar/na
