🧠 Why Hasn’t AI Increased Unemployment?

Sign Up | Home Page | Meet The Team

In today’s sunday special:

Read time: 7 minutes

📜 THE PRELUDE

Given recent headlines, AI certainly looks like a job killer. Big Tech companies have slashed nearly 140,000 jobs since the start of 2026, claiming the layoffs are to prioritize “AI-driven growth strategies.” But is this true?

What we’re witnessing isn’t an unbiased interpretation of reality, but rather an incentive-driven narrative. Capital benefits from optimism. Labor benefits from caution. The Media benefits from attention. This doesn’t mean concerns are fake. It just means the loudest claims aren’t necessarily the most accurate ones. Today’s polarizing narrative that “your job will be replaced by AI” is driven by marketing hype and media sensationalism rather than technical reality.

💼 EDUCATED BUT UNEMPLOYED

The U.S. economy has historically linked corporate profits with human payrolls. When enterprises earn more money, they typically respond by hiring more employees. It’s the most common way to eliminate capacity constraints and sustainably scale. For example, if a boutique consulting firm signs 25% more clients, it hires additional consultants to maintain the same high level of client service across its growing client portfolio.

Anthropic CEO Dario Amodei famously warned of a radical economic “decoupling”: a future where GenAI productivity gains yield explosive annual GDP growth of up to 10%, while simultaneously driving unemployment levels as high as 20%. In simple terms, less human labor is needed to create wealth because GenAI can automate content writing, data analysis, and code generation. He even went so far as to predict that GenAI could erase half of all entry-level white-collar jobs within one to five years. He coined it the “white-collar bloodbath.”

Imagine a modern law firm that leverages AI Agents: software programs that analyze, arrange, and automate on your behalf without you lifting a finger. These AI Agents complete the everyday caseloads of 50 paralegals, making the modern law firm 100x more productive while reducing headcount to a skeleton crew of a few managing attorneys. To address this AI-induced job displacement, Amodei controversially suggested a “Token Tax” on AI-generated revenue to redistribute wealth. Instead of paying 50 paralegals, the modern law firm would pay taxes on the tokens: the “digital units” of work generated by AI Agents. The U.S. government would collect roughly 3% of every token generated and funnel it into a fund to pay for UBI.

Anthropic recently introduced a new hybrid metric called “observed exposure,” which measures how frequently different industries across the U.S. economy are actually employing AI to automate, not assist with, economically valuable tasks. They identified that computer programmers have the worst observed exposure at 74.5%. While computer programmers face the highest potential job loss, they’re not necessarily the ones losing their jobs right now. In other words, the 74.5% represents the “front door” of this career path closing.

According to Oxford Economics, the rate of unemployment among recent U.S. college graduates aged 22 to 27 is nearing 6%, which is higher than the national unemployment rate of 4.2%. “It’s the first time this has happened in the last 45 years,” explained Matthew Martin, Senior U.S. Economist at Oxford Economics. “There’s a mismatch between business demand and the labor supply overall, and it’s very concentrated in the technology sector.” For context, U.S. workers aged 22 to 25 in “AI-exposed occupations” have experienced a 16% decrease in job finding rates.

For U.S. college graduates, entry-level white-collar jobs serve as the launchpad for successful careers. If they’re denied that professional development, it can drag down the U.S. economy by reducing consumer spending, postponing family formation, and delaying homeownership. Amodei’s partially right that we’re experiencing an entry-level job squeeze right now, but is GenAI to blame?

🥩 HUMAN UTILITY ALWAYS WINS

Amodei’s doomsday employment diagnosis sparks dialogue, but it arguably ignores the historical evolution of human utility. Approximately 60% of U.S. workers today are employed in job roles that didn’t even exist in 1940. In the last 80 years, more than 85% of job growth has been driven by new technologies. “Predictions of technology reducing the need for human labor have a long history but a poor track record,” said Macro Research Analyst at Goldman Sachs Sarah Dong.

Since the rise of Industrial America, U.S. workers have feared that machines would replace their skills and make their labor obsolete. In reality, when machinery transformed traditional industries like food processing, it didn’t lead to massive job cuts. Instead, job roles were broken into simpler, more repetitive steps. For example, large assembly lines with specialized sections such as slicing, trimming, and packaging replaced skilled butchers in small meat shops. The meatpacking laborers complained of “speed-up, work intensification, and work degradation,” as described by labor historian Jason Resnikoff.

AI adoption is spreading at a historical speed. AI reached 53% population penetration within three years, scaling faster than the internet or the PC revolution. It’s the fastest-adopted and fastest-growing general-purpose technology in human history.

There’s more and more money flowing into AI. For context, global corporate AI investments reached $581.7 billion in 2025, up 130% from the prior year. Meanwhile, regional private investments reached $344.7 billion in 2025, up 128% from the prior year. The U.S. leads all other countries in “doing out AI dollars,” investing $285.9 billion, or 23.1x more money, than the next leading country.

In an interview with 1,000 executives, 98% anticipate “near-term organizational design changes over the next two years.” According to BCG, 50% to 55% of job roles in the U.S. will be reshaped by AI within the next three years, while 10% to 15% could be eliminated within the next five years. AI adoption is moving 6x faster than the internet did across enterprises.

Despite this, human utility still wins. OpenAI CEO Sam Altman initially claimed that AI would replace “most of the jobs people do today” and that entire job categories would be “totally, totally gone.” Now, he believes the “AI jobpocalypse” won’t happen because the human part of employment matters more than he thought.

Rather than triggering mass unemployment, Amazon founder Jeff Bezos argues GenAI will lead to labor shortages by accelerating the “dream-build” loop. We constantly dream up endless inventions, but execution has always been the bottleneck. As GenAI expands our capacity to execute, the supply of inventions will outpace the supply of labor.

The latest U.S. private employment data suggests AI isn’t the ultimate job killer. Instead, it’s the ultimate job re-skiller that’s creating more work with better wages. According to Torsten Slok, Partner and Chief Economist at Apollo, there’s “zero evidence of AI-related job losses.” More specifically, he cited the ADP National Employment Report, which analyzes weekly payroll data from more than 26 million U.S. employees to track changes in U.S. private employment. In April 2026, U.S. private employment increased by 109,000 jobs, with annual pay for “job-stayers” up 4.4% and annual pay for “job-changers” up 6.6% from the previous year. It’s the best month for U.S. job growth since January 2025. In other words, the U.S. labor market is adding more jobs with higher wages.

The LFPR sits at 61.5%, meaning six in ten working-age Americans are either employed or actively seeking employment. Although the LFPR remains well below the 67.3% peak reached in 2000, it’s attributed to the long-term decline of Baby Boomers aging out of the U.S. workforce, while Millennials and Gen Z prioritize education over employment. Meanwhile, the UR sits at 4.2%, comfortably below the 50-year average of 6.1%. More importantly, U.S. employment is expected to grow 4.0% by 2033, indicating a steady, long-term expansion of the U.S. job market.

So, why does the U.S. job market seem so bad right now? We’re experiencing a low-hire, low-fire stagnation that feels brutal to job seekers despite the 4.2% UR. Big Tech companies have defaulted to “AI-washing,” rebranding job cuts caused by overstaffing from the post-pandemic boom as strategic AI pivots to boost investor sentiment and inflate stock prices. According to CGC, 1,206,374 job cuts were announced in 2025. Of those, only 54,836, or about 4.5%, were explicitly attributed to AI.

💪 IT CREATES MORE WORK, NOT LESS

In 1967, American computer architect Gene Amdahl pioneered “Amdahl’s Law”: when you speed up one part of the process, whatever you can’t speed up becomes the new choke point. Consider the airport check-in process. You can install high-speed check-in kiosks with automated bag drops, but if only a few TSA officers are available to check IDs, lines will form at that stage. The slowest part dictates the pace of the entire system.

Paralegals often organize discovery, prepare exhibits, and summarize case law. Harvey, AI software for legal and professional services, can synthesize testimonies and summarize depositions within minutes. What once took paralegals hours to complete can be spun up instantly. At first glance, this seems like a pure productivity win, but the choke point simply shifts. Every piece of legal work generated by Harvey must still be reviewed by a paralegal. If Harvey introduces a subtle legal error, the responsibility rests entirely on the modern law firm.

GenAI doesn’t increase operational efficiency when it amplifies existing constraints. In this instance, the ability to synthesize testimonies and summarize depositions within minutes results in more casework for paralegals to review for basic legal accuracy. The choke point simply shifts from drafting to reviewing.

In 1865, English economist William Stanley Jevons noticed something surprising. When engineers invented more efficient engines that relied on less coal, Britain’s total coal consumption increased. He coined this phenomenon the “Rebound Effect,” where increased efficiency of a resource doesn’t decrease total consumption but rather increases it. We see this phenomenon in human behavior today. People who drive fuel-efficient vehicles often drive farther than they otherwise would because they feel like they’re “getting more per gallon.” This mindset can reduce the expected fuel savings from technological advances in transportation by as much as 30%.

After analyzing 200 employees at a U.S.-based technology company for eight months, HBR discovered that AI consistently “broadened the job scope,” promoting a faster work pace and extending the workday. The U.S.-based technology company didn’t mandate the adoption of AI. On their own initiative, the 200 employees did more because AI made doing more “feel possible, accessible, and intrinsically rewarding.” But when the novelty of their newfound superpowers wore off, they reported “burnout” and “cognitive fatigue.”

When conducting 40 in-depth interviews across the design, product, and engineering teams, each team member consistently conveyed that AI “blurred boundaries between work and non-work.” Since AI made beginning a simple daily task so easy, team members would slip small amounts of work into lunch breaks. Over time, it fostered a workday with “fewer natural pauses” and “more continuous involvement.” Even though AI reduced the mental effort required to complete a simple daily task, they didn’t just finish the simple daily task faster and move on. Instead, they ended up doing more of them.

🔑 KEY TAKEAWAY

AI exemplifies a skill-biased, labor-augmenting technology. It complements human expertise and broadens the scope of human capability. Rather than reducing the need for human labor, it shifts where human value is created, with the burden of judging, refining, and applying more economically valuable than ever.

📒 FINAL NOTE

{{rp_personalized_text}}

Share your unique referral link: {{rp_refer_url}}