India’s declining fertility threatens China-like demographic crisis, says Sanjeev Sanyal

The decline in fertility is no longer limited to the southern states but has emerged as a nationwide trend, with the country's total fertility rate (TFR) dropping below the replacement level across several regions, Sanyal told Mint in an interview. India, according to him, is now entering the same demographic transition zone that has challenged countries such as China, Japan and South Korea.

“If it weren’t for Bihar and Uttar Pradesh, we would be well below the replacement rate. Even now, we are already below replacement levels, and without those states, the situation would be far worse," he said. “This is a serious national concern that requires urgent attention."

Also read: Demand for MNREGS jobs up, but may decline soon

The total fertility rate in India, the world’s most populous country, has declined to 1.9 births per woman, falling below the replacement level of 2.1, according to the latest UNFPA's 2025 State of World Population report. Political leaders in the southern states have begun advocating for policy incentives, such as tax breaks, cash benefits and childcare support to encourage larger families.

“Oddly enough, we still have population control departments operating across the country. These are outdated and must be shut down, even in Bihar and Uttar Pradesh," Sanyal said. “Not only are birth rates declining in those states, too, but we are increasingly reliant on them to keep our national fertility levels at a reasonable threshold."

According to Sanyal, India's population peaked in annual births nearly 15 years ago, and declining student numbers, especially in southern and Himalayan states, are now forcing schools to shut, making the old notion of a population explosion increasingly outdated.

“Whenever I mention school closures, it tends to trigger emotional responses. But the reality is, in many regions, now more than half the country, schools simply don’t have enough children to remain viable," he said. “Classrooms are shrinking, and resources are being spread thin."

‘Do not attempt blanket AI regulation’   

Even as artificial intelligence (AI) disrupts businesses, jobs and society, Sanyal is against the global push for blanket AI regulation, arguing that a one-size-fits-all approach would stifle innovation.

As governments around the world grapple with how to regulate AI, Sanyal said two dominant models have emerged. The American model leans toward self-regulation, with companies largely free to innovate while the state intervenes only when problems surface. This system, he argues, allows for faster technological progress and market-driven adaptation.

In contrast, the European model seeks to regulate AI preemptively, relying on bureaucrats to forecast potential harms and impose safeguards in advance.

India should strike a balance between enabling innovation and ensuring responsible deployment, a regulatory philosophy that diverges from the sweeping frameworks being explored in the European Union and elsewhere, he said.

“In India’s case, I have proposed a completely different model — one that comes with its own trade-offs. My suggestion is: do not attempt a universal AI regulation. Instead, deliberately compartmentalize AI systems from the outset," Sanyal said.  

Also read: PM-EAC member Sanjeev Sanyal proposes a CAS-based framework to regulate AI

“There should be strict silos—for example, the AI that runs banking systems must be completely disconnected from the AI running satellites or the electricity grid," he added.

In a research paper last year, Sanyal proposed a flexible AI regulatory framework, calling for a specialist regulator, a national algorithm registry, and a centralized repository to drive innovation.  

Co-authored with Pranav Sharma and Chirag Dudani, the paper outlined a model based on complex adaptive systems, emphasizing responsive, evolving oversight over rigid controls.

This includes five key principles, which include setting guardrails to contain harmful AI behaviour and ensuring manual overrides and chokepoints to keep critical infrastructure under human control. In the paper, Sanyal argues that AI regulation must adapt in real-time, not rely on static rulebooks.

Also read: India's unemployment rate at 5.1% in April; labour force participation at 55.6%

“So, my first recommendation is to create hard boundaries between AI systems, even if that causes some inefficiency. Everything does not need to be connected to everything else," he said. “Second, there must be legal requirements for human overrides at every critical control point. And third, and most important, we must institute explainability audits."