Transcription
Sarah graduated top of her class with a finance degree. She applied to 247 entry-level jobs and got zero callbacks. Her roommate, a paralegal, was laid off - replaced by an AI that works for pennies. And she’s not the only.
Hi, Josh here. And today on The Infographics Show, we’re explaining how AI is causing a white collar purge. In May 2025, Anthropic CEO Dario Amodei warned that AI could wipe out nearly half of entry-level white-collar jobs and push unemployment to 10–20% within five years. It’s a terrifying thought - especially when it’s coming from the guy that’s building the technology. That’s not some far-off science fiction. It’s reality.
Amodei went so far as to claim the government should “stop ‘sugar coating’ what’s coming, namely, ‘the possible mass elimination of jobs across technology, finance, law, consulting, and other white-collar professions, especially entry-level gigs.” And the question is - will anyone stop it?
Ford CEO Jim Farley predicted AI would halve the number of white collar jobs. When the CEOs of Fortune 500 companies start throwing around phrases like “bloodbath” and “tsunami,” you know something is up. And the numbers? They back it up. This whole thing is still playing out, but early returns are eye-opening to say the least.
In the first six months of 2025, almost 78,000 tech job losses were directly attributed to AI. That’s over 400 highly qualified people losing their job… every single day. In January 2025, the U.S. Bureau of Labor Statistics reported the lowest rate of job openings in professional services since 2013, a 20% year-over-year drop. The bleeding isn’t theoretical.
Based on an analysis of 180 million job postings, computer graphic artists fell 33% in 2025 after dropping 12% in 2024. Corporate compliance specialists dropped 29% in 2025. These are college-educated professionals with specialized skills whose prospects are melting away, not factory workers being replaced by robots. Remember when your parents told you to get a college degree for job security? We should probably talk about that.
McKinsey, a leading consulting firm, estimated that generative AI could automate activities that currently represent between 60 and 70 % of the working time in certain office occupations. Need a translation? Most of what entry-level workers can do can now be done by software… apparently.
Customer service, naturally, was the canary in the coal mine. AI-powered chatbots already handle more than 85% of first-level support requests in large tech companies - up from 30% in 2020. Experts are predicting that number to rise to 90% by 2029, a staggering number. In just five years, the human touch is nearly gone. And it’s not stopping there.
Salesforce CEO Marc Benioff confirmed his company cut another 4,000 customer service positions following the rapid integration of AI “agents” handling roughly 50% of customer interactions. Imagine walking into your first office job and discovering half your coworkers are software.
Law firms might have thought they were safe behind their degrees and expensive suits. Well, they’re not. Tools like CaseText conduct legal research and draft contracts in minutes, a task that previously took a junior lawyer hours. Things are getting dire. Paralegals are now facing an 80% risk of automation by the end of 2026. Remember how lawyers used to bill $300 an hour for a first-year associate to review documents? AI will do a hundred thousand times more output for three bucks. Guess who’s winning that margin call?
And the scale of change is staggering. The World Economic Forum projected that by 2027, approximately 83 million jobs could be lost globally due to AI and automation. A Goldman-Sachs report is estimating that 300 million full-time jobs worldwide could see their jobs affected in some way by generative AI. In the U.S. specifically, up to 46% of tasks performed by entry-level employees could be automated in the next decade. That translates to roughly 10-12 million entry-level office jobs possibly vanishing overnight.
Want to know what’s really happening? Just listen to what companies are actually doing, not what they’re saying in press releases. A few years ago, IBM’s CEO was stating that the company planned to halt hiring in almost 8,000 roles that could be replaced by AI. Human resources and administrative support were in the cross hairs - representing about 26,000 employees. In 2023, Accenture announced the cut of 19,000 jobs, mainly among non-billable and entry-level staff, while increasing its investment in training and deploying generative AI.
Amazon is the latest to enter the entry-level apocalypse. They announced 16,000 corporate job cuts - half of the 30,000 eliminated since October, all in the name of AI and efficiency. For a company with 1.5 million employees, that may sound small - until you realize it’s 10% of the corporate hierarchy. Notice the pattern? Fire the humans, invest in the robots. It’s not even subtle anymore.
JPMorgan’s managers have been told to avoid hiring people as the firm deploys AI across its businesses. That’s one of the world’s largest banks essentially saying, “We’re good on humans, thanks.” And things can get even more dystopian.
If you’re a college undergraduate, now might be a good time to plug your ears. Internships, yes, that sacred American institution of getting coffee and making PowerPoints for college credit, is dying. Listings have fallen by more than 15% from January 2023 to January 2025. At the same time, applications have dramatically increased. The decline is even more severe in high-paying fields. Technology postings dropped by 30%, and professional services postings dropped by 42%.
According to career site Handshake’s 2025 Internship Index, the average applications per internship jumped to 109 for 2024–25, up from 62 the year before and 43 in 2022–23. That means internships are seeing around two-and-a-half times as many applicants as they did just 2 years ago. Internships in the tech sector were the most competitive with 273 applications per posting. That was followed by financial services at 192 and professional services at 187. Internships are now a battlefield.
The desperation has gotten so intense that AI and tough entry-level job markets are making it common for students to begin applying for internships as freshmen or sophomores. Why hire interns when AI can do the work? Companies are finding that young workers once essential for tasks like transcription, data labeling, or content moderation are no longer indispensable.
And for those lucky few who actually land jobs? Things aren’t looking good. According to the Federal Reserve Bank of San Francisco, real wage growth in entry-level professional jobs fell by 1.8% in 2024. That’s in contrast to a 2.4% increase across the overall economy. So while everyone else is getting raises, entry-level workers are getting pay cuts. Because when there are 273 applicants for every tech internship, why pay more? Welcome to the new reality.
And the math, in general, is just brutal. A 2024 MIT study found that the adoption of AI led to a 19% drop in job postings involving repetitive cognitive tasks. But for every new job created in AI, two to three junior jobs disappear or are never posted. So AI creates one $150,000 machine learning engineer position while eliminating three $50,000 entry-level analyst jobs. Sure, the total payroll might be similar, but tell that to the two people who never got hired in the first place.
Every time someone predicts a job apocalypse, skeptics trot out the same argument. “They said this about the printing press, the steam engine, the computer!” Time and again, technologies that were supposed to wipe out whole categories of work ended up doing something quieter and stranger. They actually changed how people worked, without erasing the work itself. Jobs adapted, and roles shifted. New tasks appeared alongside old ones. Entire professions reorganized themselves around the new tools rather than being crushed by them. What survived was the work that required context, nuance, accountability, interpretation, and trust… the parts that machines are worst at owning.
But what if this time is different? Whereas previous waves of automation such as spreadsheets supported human work, generative AI can perform entire tasks and workflows from start to finish without or with minimal human intervention. Spreadsheets didn’t do your accounting for you; they merely helped you do the accounting faster. AI actually does the accounting. That’s the leap.
Anthropic’s research shows AI is currently being used mainly to help people do their jobs - a process known as augmentation. This will increasingly shift towards AI actually doing the jobs themselves - automation - in as little as a couple of years. Thinking we can talk our boss out of integrating a bunch of glorified chatbots? That trick may soon be over.
AI agents - software programs that act autonomously to understand, plan, and execute tasks - are getting smarter by the day. They interface with existing tools and system networks as needed to fulfill user goals. These things don’t need bathroom or coffee breaks, health insurance or performance reviews. They just… work.
But what happens when an entire generation can’t get entry-level jobs anymore? They’re about to become another casualty in an economic disaster unfolding in real time. AI could destroy the very architecture of professional careers. Entry-level positions matter. They teach skills, enable upward mobility, and keep the pipeline of talent flowing to mid and senior-level roles. Remove them, and the future workforce collapses.
Companies need senior employees. Senior employees with experience, judgment, and institutional knowledge. But those people don’t just materialize out of thin air. They start as nervous 22-year-olds making mistakes on small projects, learning the ropes by doing and building key skills over the years. If you eliminate all the junior positions, where do future senior employees come from? It’s like removing the minor leagues and wondering why you are not producing the same caliber of major league players in 10 years.
And the immediate economic consequences are spreading fast. And you need to brace yourself for the fallout. Look no further than the student loan repayment predicament. As of December 2025, 1 in 3 borrowers are making payments more than 90 days late. One in five have stopped paying altogether. A projection from the Congressional Budget Office warned that if the current trend continues, student loan delinquency could rise even more. We are already on the precipice of what many are calling the “default cliff” - young people struggling to keep up and staring into the abyss.
Students are told education is an investment in their future. Now that “investment” is actively shrinking their job prospects - as entry-level finance roles are automated faster than they can even apply. The unemployment rates for people aged 22-27 with at least a Bachelor's degree is consistently higher than the national rate. We’ve created a system where the more educated you are, the harder it is to find work. College is becoming a luxury purchase that decreases your employment prospects. That’s not how it was supposed to work.
And the ripple effects compound. Unemployed graduates don't buy houses. They don't start families. They don't consume at the levels economists projected when they took out those student loans. They move back in with parents, delay major life decisions, and watch their savings evaporate - all while AI does the jobs they trained for.
There are a few subset career paths that may well survive this AI-white collar purge. While software engineers have every reason to panic, plumbers and electricians are chilling. The data backs this up. Parents overwhelmingly point to trades as safer than digital roles; these roles are seen as AI-resilient, compared to their tech-field peers. And if you are entering the tech field, make sure you are the one learning to build the datacenters, not run them. Just ask Nvidia CEO Jensen Huang, who recently told the World Economic Forum that he foresaw “a lot” of six-figure jobs being unlocked as part of the unending AI data center buildout across the globe. It’s part of what he calls the greatest infrastructure project humanity has ever undertaken. Meanwhile, the United States is still short of hundreds of thousands of factory workers, construction workers, and auto technicians who remain in sharp demand. Turns out the real job security was the manual labor we looked down on along the way: The electrician who fixes the server farm has better prospects than the paralegal who got replaced by the server farm.
So what is actually happening? Alarmists see a white-collar apocalypse. Skeptics point to past false alarms. But realists know most experts expect shrinking growth - or outright losses - by 2030. Still, unemployment has not spiked, and data is consistent with either a normal technological transition or the early stages of something far more disruptive. Which path will we be on in five years? Whether this becomes a catastrophe or just another economic transition depends on choices being made now about how we deploy this technology. How we distribute the gains, and what we do about the people left behind.
And Sarah is still out there, 247 applications deep, watching AI chatbots send her automated rejection emails after one scan of her resume for jobs AI will eventually do entirely anyway. Her friend who went to trade school? She’s got three job offers and is currently negotiating with the one which pays the most. Guess who’s winning in this new economy. Now go watch AI Has Already Won, You Just Can't See It Yet. Or click on this video instead.