The End of AI Is Electrician Work, This Image Has Silicon Valley People Losing It

The weekend just passed, and Meta’s 16,000 employees may need to find new jobs.

Reuters, citing insiders, reports that this round of layoffs could affect over 20% of the company’s staff, ostensibly to offset huge investments in AI infrastructure.

Meanwhile, OpenAI co-founder Andrej Karpathy has systematically ranked all U.S. jobs based on their exposure to artificial intelligence.

But for some reason, the link was inaccessible less than ten minutes after release. (It is now back online.)

He extracted 342 occupations from the U.S. Bureau of Labor Statistics, scored each position with an LLM, and created a tree diagram to visually show the size and AI risk level of each profession.

URL link: 🚪

What we see is: among 143 million U.S. workers, the weighted average exposure score is as high as 4.9 out of 10; over 25 million people (18%) are in the “very high risk” zone (8-10 points), and another 34.7 million (24%) fall into the “high risk” zone (6-7 points).

In other words, nearly 40% of the U.S. workforce is directly facing the wave of AI replacement.

Blue-collar and white-collar, thirty years east, thirty years west

When artificial intelligence was just a concept in science fiction, people worried about machines replacing factory line workers. But today, the trend is the opposite.

Based on threat levels, current jobs can be divided into several tiers.

Very high risk zone (8–10 points):

• Customer Service Representatives (9/10, 2.8 million)
• Secretaries and Administrative Assistants (8/10, 3.5 million)
• Office Clerks (9/10, 2.6 million)
• Bookkeepers (9/10, 1.6 million)
• Financial Clerks (9/10, 1.2 million)
• Software Developers (9/10, 1.9 million)
• Computer Systems Analysts (8/10, 520,000), Technical Support (8/10, 880,000)
• Accountants and Auditors (8/10, 1.6 million), Market Research Analysts (9/10, 940,000)
• Lawyers (8/10, 860,000), Paralegals (high risk)
• Bank Tellers, Insurance Underwriters, Securities Brokers

Interestingly, Anthropic’s own labor report also lists these occupations, with computer programmers exposed at 74.5%, customer service reps at 70.1%, data entry clerks at 67.1%, and medical records specialists at 66.7%.

High risk zone (6–7 points):

• Cashiers (7/10, 3.2 million)
• Financial Managers (7/10, 870,000), Sales Managers (7/10, 620,000)
• Educators: High school teachers (7/10, 1.1 million), University Professors (7/10, 1.4 million), Elementary school teachers (6/10, 1.5 million)
• Executives (6/10, 4 million)
• Doctors and Surgeons (5/10, 840,000)
• Heavy truck drivers (5/10, 2.2 million)
• Security Guards (5/10, 1.3 million)

Medium risk zone (4–5 points):

• Registered Nurses (4/10, 3.4 million)
• Retail Salespersons (4/10, 4.2 million)
• Farmers and Agricultural Managers (4/10, 840,000)
• Teacher Assistants (4/10, 1.4 million), Medical Assistants (4/10, 810,000)
• Police Officers and Detectives (4/10, 830,000), Social Workers (4/10, 810,000)

Low risk zone (0–3 points):

• Home Health Aides and Personal Care Aides (2/10, 4.3 million)
• Cleaners and Janitors (1/10, 2.4 million), Grounds Maintenance Workers (1/10, 1.3 million)
• Construction Workers (1/10, 1.6 million), Movers (2/10, 7 million)
• Carpenters (2/10, 960,000), Electricians (2/10, 820,000): tasks requiring fine manual skills and non-standard circuit repairs
• Chefs (3/10, 2.8 million), Food Service Workers (3/10, 5 million)
• Barbers and Beauticians (2/10, 650,000)
• Auto Mechanics (3/10, 810,000)

Jobs earning over $100,000 have an average exposure score of 6.7, while low-income jobs earning less than $35,000 average only 3.4. Over the past thirty years, “brain workers” in the knowledge economy have benefited most—receiving better pay and higher status—while those who earn a living with their bodies have been seen as prime candidates for automation. Now, the trend has reversed.

Anthropic’s research team calls this trend the “White-Collar Great Recession” in 2025. The report states that if the unemployment rate in high-exposure AI jobs doubles from 3% to 6%, it would be comparable to the doubling of overall unemployment during the 2007–2009 financial crisis.

Just last weekend, Reuters cited insiders saying that Meta is planning large-scale layoffs, potentially affecting over 20% of its staff. The move aims to offset massive investments in AI infrastructure and prepare for increased efficiency from “AI-assisted employees.” With nearly 79,000 employees, this could mean about 16,000 jobs cut.

Mark Zuckerberg said in January this year, “Projects that once required multiple large teams can now be completed by a single talented individual.” Many people at the time saw this as a praise of productivity. Looking back now, it can also be seen as a prelude to layoffs.

Since early 2025 through the end of the year, U.S. companies have laid off over 1.17 million workers, the sixth time since 1993 that this number has been exceeded.

Is the end of AI the electrician?

Regarding Karpathy’s tree diagram, some suggest taking a different perspective: “Screen-based jobs score highest in AI exposure, but these are also the roles most suitable for integrating AI into workflows. These scores measure vulnerability, not adaptability. A software engineer scoring 9 doesn’t necessarily mean they’ll be laid off; perhaps they’re producing three times as much.”

But if a company decides to cut a team from 10 to 3 people, what about the other 7? Regardless of their ability to adapt to AI, their immediate concern is: where to find the next job.

Currently, one of the most stable options for those 7 people might really be becoming electricians.

According to the latest data from the U.S. Bureau of Labor Statistics, the median annual salary for electricians is $62,000, with a projected growth rate of 9% over the next decade—much faster than the average for all occupations. Ian Andrews, head of the National Electrical Contractors Association, states: “We lose about 20,000 retiring electricians each year, and there are 80,000 job openings to fill.”

What drives this demand? AI itself. Nvidia CEO Jensen Huang has said that building data centers for AI requires a large number of electricians, plumbers, and HVAC technicians. He predicts “every economy’s technical trades sector will experience explosive growth.” A 250,000-square-foot data center during construction can employ up to 1,500 workers, many earning over $100,000 annually, without requiring a college degree.

Interestingly, a generation once “delayed” by vocational education is now reversing course. The International Brotherhood of Electrical Workers (IBEW) describes the electrician shortage as a “life-and-death” crisis, while more Generation Z individuals are reconsidering trades like electrical work, HVAC, and plumbing.

Of course, this path isn’t without its cracks.

Some netizens added robot factors to Karpathy’s original chart, making the diagram even redder. Roofers, cleaners, construction workers, plumbers, electricians—they face a different set of automation threats: robotics, autonomous driving, and warehouse automation are gradually encroaching.

In other words, once robots with human-like mobility become technically feasible, jobs currently protected because they require “on-site presence” will face a new wave of disruption.

Today’s safe zones may not be tomorrow’s.