Former RBI governor Raghuram Rajan has proposed an “AI tax” as one possible response to the risk that artificial intelligence could displace large numbers of workers. His argument is that existing tax systems can unintentionally make machines cheaper than human employees: companies pay payroll-related contributions for workers, while there is no comparable charge when AI performs the same work. Rajan suggests taxing the AI tokens companies consume as a way of partially balancing that incentive.
Rajan does not argue that AI adoption should be stopped. Instead, he says the pace and scale of adoption will determine how disruptive the technology becomes. Current business adoption remains far from universal, with smaller companies lagging larger firms, while many enterprises are still experimenting because integrating AI into existing workflows and determining its costs remain difficult. But competitive pressure could eventually accelerate adoption and therefore increase the risk of rapid labour displacement.
His proposal would begin with a relatively low levy on AI-token usage, increasing gradually as governments gain experience, while carefully calibrating the tax so it does not unnecessarily discourage productive AI deployment. Rajan also proposes tax credits for companies that retrain and retain workers, potentially tying the credit to how long those employees remain employed. The objective would be to make workforce adaptation part of the economics of AI adoption rather than leaving displaced workers to bear the transition alone.
Importantly, Rajan's outlook is not entirely pessimistic. He argues that AI could create new jobs, make existing workers more productive and lower the cost of starting businesses, potentially generating additional employment. The central policy challenge, therefore, is not simply whether AI destroys jobs, but how societies manage the transition between old and new forms of work. His proposed AI tax is consequently less about punishing technology than about correcting the economic incentives around automation and funding the human side of the transition.
