The Survey, a pre-Budget report that looks at how the economy performed over the past year, reviews key government programmes and outlines policy moves. It also gives a snapshot of where the economy is headed in the near future. Here's what Chapter 13 of the document, ‘Labour in the AI era: Crisis or catalyst?’, says.

AI Job Cuts Could Mean Higher Taxes

The Survey says that in a labour-rich nation like India, with its huge population and relatively low per capita income, the impact of AI on jobs will be felt in every sector.

This is a global phenomenon and not unique to India. The International Labour Organisation (ILO) has estimated that across the world, around 75 million jobs could be replaced by AI, specifically among clerical support workers.

Goldman Sachs puts that figure at 300 million full-time jobs in the private sector. McKinsey, on the other hand, says that by 2030, up to 30 per cent of current work hours could be automated by generative AI across Europe and the US.

The Survey categorically says the onus of job losses from AI and the proper implementation of the technology falls on companies. It says that if companies don’t roll out AI carefully, or if there are job losses as a result of deploying AI, then government intervention in the form of policy will become a necessity.

It is possible that the government will have to tax the profits companies are expected to make as a result of replacing human workers with technology. “It will leave everyone worse off and the country’s growth potential will suffer as a result,” noted the Survey.

In order to bridge the gap between the growing demand for AI expertise and the current skill-set of the workforce, top IT companies in India are training their employees in generative AI. These companies recognise the importance of upskilling their employees to ensure they remain competitive and capable of handling the complexities of AI-driven tasks.

“India is showing the way when it comes to integrating AI in the future of work,” said President Droupadi Murmu in her address in Parliament today.

The Survey suggests that while predictions about how much AI could affect jobs may be exaggerated, it’s still dangerous for India to ignore the potential risks. It says that even though the chance of a big issue — like massive layoffs — cropping up, might seem low, the consequences could be so severe that it's important to take it seriously and prepare for it.

Resource And Infrastructure Challenge

The Survey says that India faces a resource and infrastructure challenge in the field of AI. In terms of building a competitive AI infrastructure, India’s job is cut out. While building data centres and chip supplies is crucial, it's not just about the physical tech.

The Survey notes that the real challenge lay in collecting and cleaning data, which is essential for training AI models. This process is time-consuming and prone to biases. The problem? India is still in the early stages of building this support infrastructure.

To add to this, India still doesn't have a data privacy law. On January 3, 2025, around 17 months after the Digital Personal Data Protection Act was signed into law, the government released the rules that provide necessary details and an actionable framework to strengthen the legalese. Currently at the stage of collecting public feedback for its rules, the digital privacy law still has a long way to go.

As of now, the government’s focus is on assuring affordable AI infrastructure to startups, researchers and academia. Yesterday, IT Minister Ashwini Vaishnaw called for proposals to build India’s own foundation model, which will be powered by the India AI Compute Facility with over 18,000 GPUs.

Researchers, startups and MSMEs will soon have access to these GPUs for AI projects, with 15,000 of these being high-end models offered at discounted prices of up to 47 per cent. 

This comes in the backdrop of Chinese company DeepSeek releasing a foundation model that performs better or on a par with models that power ChatGPT and Claude chatbots.

What sets it apart, and has caused a furore globally, is that the Chinese model was built using 1/10th of the GPUs used by big tech companies, that too, within the span of two months. It challenged the long-held belief that AI models can only be developed with thousands of advanced AI chips and massive computational power.

Since India currently lacks both the appropriate infrastructure and resources, the promise of a strong Budget backing for India's own foundation AI model has infused the industry with some hope.

Coming to resources — financial or human — the Survey notes that it comes at a cost. Data centres, which house GPUs and power AI, devour electricity and water. India, with its rising energy needs and resource constraints, will face tough questions: Can we really afford the environmental and social costs of rapid AI growth?

“The construction of sprawling data centres risks displacing essential land use, further exacerbating these challenges… The time to address this pressing issue is now,” says the Survey.

The Survey also lays heavy emphasis on the need for AI to align with societal values, and balancing innovation with accountability and transparency. However, it notes that "AI regulations will go only thus far".

“For India to seriously consider the effects of AI, actions will have to focus on strong institution building,” says the Survey.

The Survey then goes on to make an intriguing proposal — the establishment of stewarding institutions. These bodies would serve as custodians of emerging technologies like AI, ensuring they are applied not just cleverly, but responsibly, particularly in sectors that touch human lives most directly, like healthcare, education, etc.

By pushing for more transparency and accountability, stewarding institutions can help the public put their trust in AI and ensure that it benefits everyone, without causing harm or unfair consequences, says the Survey. 

However, it also notes that building these institutions will take time, as it is a complex task that requires intellectual and financial resources. It’s not an immediate priority, the Survey says, adding that aligning all the systems and policies to support this goal can be tough.

“...whether the world is better or worse off with AI is not easily answered now,” the Survey concludes.