Government can bring robotics to life

This article is an on-site version of the Free Lunch newsletter. Premium subscribers can sign up to get the newsletter delivered every Thursday and Sunday. Standard subscribers can upgrade to Premium here, or explore all FT newsletters

In Isaac Asimov’s Robot science fiction series, the future is populated by sentient anthropomorphic machines. But artificial intelligence in our reality takes a different form. Like HAL, the AI in Stanley Kubrick’s 2001: A Space Odyssey, large language models are not embodied. They move bits through the economy, not atoms.

There are reasons to want Asimov’s robot-filled future. Machines can replace physically taxing and dangerous labour. Robotics can unlock growth in previously stagnant sectors. Automation in the physical world can free up people to find more interesting and meaningful tasks.

Just as the agricultural revolution liberated large swaths of humanity from back-breaking farm work, automation could improve livelihoods. But such a future is not inevitable, especially for sectors that rely on cheap physical labour.

Some of the biggest challenges for automation in the physical world are economic. For many industries, capital-intensive robots do not make much sense from a cost perspective. Building models and data centres that make coders more productive can be lucrative; building robot janitors less so. And investment in technologies that replace high-value knowledge-economy tasks could be crowding out other areas of investment and stifling their productivity growth.

In recent decades, automation technologies have largely boosted productivity for high-skilled office workers. Massachusetts Institute of Technology economist Daron Acemoglu, whose work Free Lunch finds convincing, has illustrated how the second half of the 20th century was characterised by skill-biased technical change. In that era, employees with more education and training, such as bankers and consultants, saw tremendous productivity and wage gains thanks to new technologies. Similarly, people who today write code for a living have seen explosive productivity growth because of new AI tools, with Anthropic research suggesting Claude Code provides higher returns to users with more expertise.

That skilled workers are becoming more efficient is certainly not a bad thing. But when investors allocate scarce capital, they choose technologies with the highest returns. We should not be surprised if skill-biased technical change continues in the decades to come and investors have few incentives to automate business functions where labour costs are low. Moreover, if AI leads to a wave of “de-skilling” and Silicon Valley’s predictions of a “permanent underclass” come to life, investment in productivity-enhancing technologies for unskilled workers could decline further.

Of course, policy can prevent this. Raising minimum wages, for one, could lead to higher levels of physical automation. A 2026 study led by Erik Brynjolfsson, economics professor at Stanford University, used US data on robot exposure and imports and found that a 10 per cent higher statutory pay floor correlated with 8 per cent higher robot adoption in factories.

There are, of course, trade-offs. A higher minimum wage could lead to higher unemployment: price floors can create deadweight loss, meaning that some workers who want a job at a lower wage cannot find one. Higher minimum wages could also be inflationary, but that might not be a problem if the AI boom proves disinflationary. In any case, better unemployment insurance would soften the blow of joblessness, but at a fiscal cost.

State-led industrial policy can also bolster advanced robotics. Countries that lead in these technologies have often done so on the heels of heavy state investment. In China, robot makers receive investment and training data from the government, a deliberate approach to crowd in what the country sees as a strategic priority. Singapore and South Korea, two countries with the highest integration of robots per employee, have both relied heavily on a developmental state model for growth.

The US would be wise to follow their examples. After all, leading US tech companies innovating in advanced robotics, such as SpaceX and Anduril, have built their businesses on government contracts and grants. Tesla’s development of the Model S benefited from a $465mn loan from the US Department of Energy’s Loan Programs Office in 2010, for example. Future policies, from investment tax credits and loan programmes to funding for scientific research and development, could accelerate American innovation of automated physical-world technologies.

Washington should act as widespread anxiety about AI grips both of the two main US political parties. Technologies that free up human labour for more meaningful work, rather than replace more desirable jobs, could be popular across the aisle. There could be numerous positive externalities: higher automation in physical sectors, such as construction, could boost housing supply and lower costs, for instance.

Of course, industrial policy can crowd out other productive forms of investment. Balancing the benefits of physical automation with priorities such as growth will ultimately be a decision for voters.

Send your thoughts to [email protected].

Food for thought

Critical mineral price booms increase economic activity and local wealth around mines, but increase deforestation, according to this recent paper from Erik Katovich and Jonah Rexer.

Free Lunch on Sunday is edited by Harvey Nriapia

Recommended newsletters for you

The AI Shift — John Burn-Murdoch and Sarah O’Connor dive into how AI is transforming the world of work. Sign up

Unhedged — Robert Armstrong dissects the most important market trends and discusses how Wall Street’s best minds respond to them. Sign up