Tyler Cowen on AI, Economic Growth, and a Disorienting Transition

Tyler Cowen opens with optimism and realism. He expects AI to bring longer lives, greater wealth, entertainment and instruction, but says many people will dislike the transition and that benefits may arrive more slowly than advocates expect.

He compares the coming change with the technological transformations experienced by his grandmother, born in 1905, who saw cars, airplanes, radio and television become part of consumer life. Cowen predicts that most people's jobs will become very different from what they expected, even if some are more enjoyable or better paid.

Cowen says the uncertainty will make career advice difficult across generations. He thinks success may shift away from people who followed an established route through elite schools and professional partnerships toward people who take initiative, learn to work with agents and try different approaches.

He expects that shift to reorder status and create political and psychological strain: in his view, losing status often hurts more than gaining it helps. He points to negative reactions to AI, including among young people, as a sign that adaptation will be difficult.

Cowen predicts that AI will change most jobs without causing mass unemployment. His reasoning is that cheaper capabilities will make many more projects possible. He identifies two especially large potential categories of work: running experiments and collecting data.

In biomedicine, AI may propose many ideas that need testing in the physical world. Cowen expects people beyond those with doctorates to carry out more experiments. Some problems can be checked through simulation, he says, but many require real-world tests.

The second category is gathering information that models can use. He gives an example of recruiting people with basketball analytics expertise to help AI interpret games. Much of the world is not yet represented as data, he argues, leaving a long-running task of collecting it.

He also expects work building energy and computing capacity, whether through solar installation or other sources. Although he predicts a wealthier society with more entertainment and instruction, he acknowledges continuing scarcity and the difficulty of adjusting to different work.

Cowen says many academic colleagues dislike the prospect of changing a skill they have mastered—publishing research—into a different form of working with AI. He expects technology and health care to grow, while recognizing that individual jobs will also disappear.

Cowen offers a personal forecast that within 30 to 40 years many causes of death could be addressed, while accidents and aging may remain. He emphasizes his own uncertainty about whether he will live to see the benefits. This is his speculative outlook, not a medical prediction established by the talk.

He expects longer lives and geopolitical competition to help persuade the public of AI's value, even as people feel disoriented or demoralized. Cowen urges listeners not to overlook the potential benefits during an unsettling period.

His economic forecast is deliberately more restrained than some Silicon Valley views: he suggests AI could raise an illustrative US growth rate from about 2% to 2.5% in the short and medium term. He attributes the limit to human bottlenecks, including the difficulty of changing institutions and integrating tools.

A worker might use Claude to reduce a five-hour memo to minutes and turn the saved time into leisure. Cowen welcomes that benefit, but says an organization gains measured productivity only when it redesigns related jobs and processes so the time saved improves the wider system.

He sees this happening in coding and startups but not across much of the economy. Regulatory approval of drugs and nuclear reactors illustrates what he calls institutional stickiness: sometimes frustrating, sometimes stabilizing, and hard to combine quickly with radical technology.

Cowen argues that raising sustained growth from 2% to 2.5% could improve the trajectory of US public debt and help avoid severe fiscal adjustments. He calls AI a potential ‘plan A’ for productivity growth. The conclusion is his budget interpretation and depends on assumptions about growth, spending, revenue and interest costs.

To make a half-point increase plausible, Cowen points to US software and inventory-management improvements in the mid-1990s, which he says coincided with a roughly half-percentage-point growth increase. He argues that AI could repeat and sustain a comparable effect, while acknowledging policy and infrastructure mistakes are possible.

Cowen expects gains to be uneven. He believes poorer people, particularly in poorer countries, could benefit greatly from access to inexpensive advice, although he notes that access and awareness remain limited. He also expects initiative-taking immigrants and others willing to experiment to gain.

He predicts that some highly paid professional partners who succeeded by following established rules will lose income or status. His hypothetical move from a $2 million Manhattan role to a $350,000 energy job in Houston is intended to illustrate a painful status adjustment rather than material destitution. He worries about its politics.

Cowen criticizes higher education's focus on students using AI to cheat. In his view, if a task can be outsourced to AI, that is also a signal to reconsider what institutions teach. He estimates that large portions of the economy, including government and parts of health care and nonprofits, will adjust slowly.

He says he has shifted about two-thirds of his own time away from writing and toward mentoring, public talks, travel and work with people. Cowen expects physical presence, the ability to inspire and personal charisma to become more important, while some traditionally knowledge-centered status may decline.

On geopolitics, Cowen warns that countries without AI capabilities may become dependent on US or Chinese systems, even if open models offer alternatives. He closes with a contrast between possible long-term leisure and the immediate work and stress of keeping up with rapidly changing tools.