|
Getting your Trinity Audio player ready...
|
Not counting the pandemic, the U.S. job market last year sank to its lowest level since the Great Recession. The nation added only 181,000 jobs in 2025, down from 1.46 million in 2024.
Anemic employment numbers are continuing in 2026, with the U.S. economy shedding 92,000 jobs in February.
Many economists agree that President Trump’s tariffs and immigration policies, along with persistently high interest rates, are partially to blame, but there is increasing concern that the rapid rise of artificial intelligence is also starting to harm the employment sector.
The situation promises to worsen, and the problem is: Companies using AI to replace workers is not a bug. It’s a feature.
In fact, the potential for job losses due to AI is a significant driver of the stock market. Businesses and investors euphemistically call it “labor savings” or “reducing labor costs” – because having fewer workers theoretically makes companies more profitable and a better investment.
Goldman Sachs predicts that companies will spend more than $500 billion this year on capital expenditures for artificial intelligence and computer data centers, in part because of the potential to reduce labor costs – the single-largest expense for most businesses.
Artificial intelligence, to be sure, has tremendous potential to benefit our society, hastening advancements in health care, scientific research and clean energy. AI might even cure some forms of cancer in the next decade.
But it could also put millions of people out of work. Anthropic CEO Dario Amodei, who recently gained fame for his principled stand against the Pentagon’s demand to use Claude for autonomous weapons and domestic mass surveillance, is predicting that AI could displace 50% of entry-level white-collar jobs in the next five years.
For many investors and businesses, that would be a lot of “labor savings”. For the U.S. economy – and the people who will lose their jobs – it would be a major disaster.
If Amodei’s and similar predictions come true, about 10 million to 12 million U.S. workers could become jobless in the next half-decade, more than doubling the U.S. unemployment rate, potentially sending it above 10% – even higher than it was during the Great Recession.
Massive job losses, in turn, would have a cascading effect on our economy and society. Young workers who took on student loans expecting white-collar careers would face serious default risk, potentially triggering a consumer credit crunch. Consumer spending would also tumble, with restaurants, travel, entertainment, and retail all facing reduced demand, potentially causing even more job cuts.
That would put us on the road to a significant economic contraction and another recession. Just counting the loss of entry-level white-collar jobs alone, and assuming an average salary of $60,000–$80,000 for those positions, AI could cause roughly $720 billion to $960 billion in lost wages annually — equivalent to 3% to 4% of U.S. GDP.
Widespread unemployment can also lead to social unrest. Plus, there are concerns about long-term unemployment, since workers who can’t get entry-level job experience are unlikely to ever have professional white-collar careers.
This may sound like a “sky-is-falling”, overly pessimistic scenario. And some argue that the threat of job displacement is overblown, accusing companies of “AI washing” – blaming recent job cuts on AI when they’re actually due to other factors.
But there are warning signs that job loss due to AI is already happening. In addition to the dismal jobs numbers in the past 14 months, postings for entry-level jobs in the U.S. plummeted 35% between 2023 and 2025.
And workers are increasingly worried. Employee concern about job loss caused by AI surged from 28% in 2024 to 40% in 2026.
Anthropic predicts that AI will cause the biggest job losses in management, business and finance, office administration, computer science, engineering and architecture, social sciences, the legal profession, and arts and media.
So what can we do about the potential Jobs Armageddon? Unfortunately, it appears that the Trump administration is not inclined to put any guardrails on AI. Also, Anthropic appears to be the only major AI company concerned about job losses.
That means California must step up. In the Legislature this year, I have a bill, SB 947, that would bar companies from relying on AI to fire or discipline workers.
It’s an important first step. But it’s increasingly clear that we have more to do – perhaps by rewarding companies that hire and keep workers and use AI to improve their workers’ productivity and job satisfaction, while preventing businesses from replacing people with machines and sending our economy into a tailspin.
Editor’s note: State Senator Jerry McNerney (D-Pleasanton) authored the AI in Government Act while serving in Congress. He is also chair of the Senate Revenue and Taxation Committee and member of the Privacy, Digital Technologies, and Consumer Protection Committee.
Senator Jerry McNerney,
—30 to 35 million workers are expected to retire and exit the labor force over the next ten years. American colleges will graduate 3.3 million students every year over the next ten years (35 million student grads, 1.1 to 1.2 million graduates are international students, 59% of the international students will return to their home countries, 864,457 students will remain to get Masters degree, 203,053 will remain to get their PHD degrees). Annually, graduates will earn 2,168 Bachelor’s degrees, and 1,028 graduates will earn associate degrees. Currently, there is still an opportunity for workers aged 25 to 45 to be upskilled.
SB 947, if passed into law, will cause a mass exodus of corporations and businesses out of California.
“It’s an important first step. But it’s increasingly clear that we have more to do – perhaps by rewarding companies that hire and keep workers and use AI to improve their workers’ productivity and job satisfaction, while preventing businesses from replacing people with machines and sending our economy into a tailspin”
It’s increasingly clear that McNerney either doesn’t get it (or does not care) that California already makes it difficult and expensive to operate a business (“rewarding companies that hire”) because of the excessive regulations and taxes promoted and enacted by his own party. The “important step” that is required is to vote his party out of office to create a more business friendly environment in the state, not blame AI for a supposed “tailspin”.
Speaking of a “tailspin”, how is that “clean energy transition” working out in California with increased gasoline imports, refinery closures, and expected fuel shortages?
The science of artificial intelligence: Artificial intelligence is fundamentally a scientific effort to understand and replicate intelligent behavior using computation systems, blending computer science, mathematics, neuroscience, linguistics, and cognitive psychology.
A clear structural breakdown of the science behind AI. Grounded in what current research and what authoritative sources describe. What “AI science” really is. AI is the study of engineering of systems capable of performing tasks that normally require human intelligence-such as reasoning, learning, perception, and decision making. This includes both understanding natural intelligence and building artificial forms of it.
Core foundations of AI: Mathematics, Linear Algebra-the backbone of neural networks. Probability and statistics-modeling uncertainty, Bayesian reasoning. Optimization-training models by minimizing error functions.
Computer science: Algorithms and data structures-efficient computation. Search and planning-how agents explore actions. Complexity theory-what problems are solvable or tractable.
Cognitive science and neuroscience. Inspiration for neural networks and learning networks. Understanding perception, memory, and reasoning in humans. Liguistics-Natrural language processing (NLP) how machines interpret and generate language.
Major scientific approaches to AI: Symbolic AI (good old fashion AI)-logic, rules, and explicit knowledge representation, strength, transparency and reasoning. Limit, actions-struggles with ambiguity and perception.
Machine learning allows systems to identify patterns within data instead of relying on direct programming instructions. Includes supervised learning, unsupervised learning, reinforcement learning.
Most modern AI systems including recommendation engines and Chatbots use MI. Deep learning: Multi-layer neural networks that excel at vision, speech, and language. Powers image recognition, self-driving perception, and language models. Hybrid systems: Combine symbolic reasoning with neural networks for more robust intelligence.
Real World application of AI science: Computer vision-object detection, medical images. Natural language processing-translation, chatbots, summarization. Generative AI-creating texts, images, audio, and code.
Ethical and scientific challenges: Bias and fairness, transparency and explainability, safety and alignment, data privacy, environmental impact of large models. The future of AI science. Researchers are pushing forward. Artificial General Intelligence (AGI)-systems with broad human-like cognitive abilities. Agentic AI-autonomous systems capable of planning and acting in the world. More efficient learning