You’re Not Being Replaced by AI. You’re Being Revalued.

It was a very 2024 kind of image: a laid-off tech worker reduced to posting his availability on Manhattan streetlights with a QR code connected to his LinkedIn. “I thought that would make me stand out,” Glenn Kugelman told the Wall Street Journal.

UC Berkeley computer science professor James O’Brien immediately followed up the WSJ investigation with his own story of students with perfect 4.0 GPAs in their major contacting him worried about having zero offers. “Tech degrees no longer guarantee a job,” he wrote on LinkedIn, comparing it to the relatively recent days when Berkeley CS graduates received multiple good offers.

As a Berkeley grad myself who has worked in finance, crypto and AI, I have mentored students facing these anxieties about where to take their careers. It’s pretty clear that what is happening is very different from the boom-bust cycle of previous tech downturns.

IT sector unemployment grew from 3.9% to 5.7% in a single month earlier this year, and Mark Zuckerberg has already said AI will replace mid-level engineers in 2025. Yet we have massively overallocated our youth to roles like software engineering and the diagnostic aspects of medicine that AI will replace en masse.

This is just the start of a fundamental restructuring of what human labor is actually worth. With that in mind, we have to think about what work retains value when machines can do almost everything.

The framework

The jobs that will become more resistant to AI displacement need to be valuable, scalable, ethical and what I call “AGI-resistant.” The outcome of that last criterion is the hardest to predict.

We are potentially years, not decades, from achieving AGI — artificial general intelligence that can handle virtually any cognitive task — but the advancements toward that goal are already changing the labor market. I am concerned about a shock where large segments of the population are made unemployed very quickly.

I have serious doubts that governments would adapt quickly enough to issue universal basic income, so we need to think proactively about which roles can actually sustain people through this transition. The principle of AGI-resistant work is that there are tasks, roles and performances where humans are simply preferred, regardless of capability.

It’s present in human chess tournaments, remaining popular due to the jeopardy of error and chance, despite chess bots being definitively stronger. The Olympics limits performance-enhancing drugs to preserve natural human ability. In both cases, we are choosing the human element over optimal performance.

This preference also shows up in community work, where compassion and empathy matter and are essential for the emotional and social development of children. And in the service industry, where it’s seen as higher status to have human labor over machines. Entertainment sits at one end of this spectrum, where human preference is strongest and most durable.

Why entertainment stands out

Entertainment has proven resistant to AI alternatives so far, though sentiment can change over time. The category is actually broader than most people assume. It includes the service industry — restaurants, hospitality and personal training — where customers pay a premium for human interaction.

The durability stems from three interrelated psychological drivers: boredom, loneliness and scarcity. People crave authentic experiences such as live performances, real connection and moments that feel unrepeatable. They also want to belong to communities and feel valued. None of these can be optimally offered by AI because it’s the human element itself that is the product.

The work is valuable, scalable and offers sufficient ethical examples. This creates an odd inversion of what we have told young people for decades, pushing them toward “practical” fields like STEM degrees, medical school and corporate law. The calculus is shifting.

The scale and urgency

This isn’t a small problem we can address gradually. India’s youth population — 371 million people — faces an unemployment rate of 16%. In the United States, millions of high earners, including software engineers, will find their work replaced. Over the next three to five years, we will see major shifts as software investments reallocate and AI capabilities expand.

Any interim solution needs to employ hundreds of thousands, if not millions, of people, and it needs to happen soon. The next wave of innovation lies in the physical world. The U.S. administration should institute a new New Deal, employing millions of Americans to develop bleeding-edge public infrastructure. This could take two forms:

• Near-term terrestrial projects like national high-speed rail networks, modernized electrical grids for renewable energy, and climate-resilient infrastructure would create immediate employment while addressing urgent domestic needs.
• Longer-term space infrastructure, including launch facilities and manufacturing capacity for off-world construction, would position the U.S. for the emerging space economy.

All this work would provide large-scale employment while we navigate toward the longer-term reality where entertainment and human-centered service work become the economic foundation.

Start incentivizing now

Students should think differently about skill-building. Knowing what I know now, I would have encouraged my college self to take more probability and statistics classes. Today, quantitative literacy pairs well with the human-centered skills that will only become more relevant like understanding psychology, creating experiences people want and building genuine community.

I don’t have all the answers, but no one does. Those Berkeley students with 4.0 GPAs and zero offers deserve better than false promises about “practical” degrees. We cannot keep preparing young people for careers that won’t exist in five years.

In my view, it’s wise for us to institute the relevant legislation and large-scale solutions to incentivize a shift toward roles that are sustainable in the medium-term. I’m interested in hearing feedback and connecting with others who are working on this problem because this conversation needs to happen now, before the shock hits.

It was a very 2024 kind of image: a laid-off tech worker reduced to posting his availability on Manhattan streetlights with a QR code connected to his LinkedIn. “I thought that would make me stand out,” Glenn Kugelman told the Wall Street Journal.

UC Berkeley computer science professor James O’Brien immediately followed up the WSJ investigation with his own story of students with perfect 4.0 GPAs in their major contacting him worried about having zero offers. “Tech degrees no longer guarantee a job,” he wrote on LinkedIn, comparing it to the relatively recent days when Berkeley CS graduates received multiple good offers.