In March, Anthropic—the cutting-edge AI company behind the chatbot Claude—published a report on how AI is affecting jobs. The goal was to help us evaluate the idea that intelligent robots were about to reshape human life and make human labor unnecessary.
Back in May of last year, Anthropic’s co-founder, Dario Amodei, claimed that AI could eliminate half of all entry-level jobs within one to five years. In January, he said AI would likely become a “general substitute for human labor.” By June, he warned we might end up in “a world where the economic trade-off is stuck on hypergrowth and hyper-inequality.”
Yet Anthropic’s own report suggests that, so far, AI’s impact hasn’t lived up to those predictions. “We find no systematic increase in unemployment for highly exposed workers since late 2022,” the report stated. It also noted that the use of AI “remains a fraction of what’s possible.” For example, Claude can handle only 33% of tasks in the computer and math category, even though it could theoretically take on nearly all of them.
Forget the AI job apocalypse. AI’s real threat might be worker control and surveillance.
Spending on data centers is skyrocketing, but productivity hasn’t seen the rapid gains that tech enthusiasts’ grand predictions would suggest. In fact, labor productivity grew more slowly in the first three years of the AI era than during the information technology boom that started in the mid-1990s.
Even Sam Altman, the most visible face of AI at OpenAI, now says he doubts AI will wipe out jobs. “I don’t think we’re going to have the kind of jobs apocalypse that some companies in our space advocate or talk about,” he said in May. As MIT economist David Autor put it: “A lot of people have noticed that the world is not changing as fast as they predicted.”
This has opened up a less dramatic conversation about how AI is evolving. The emerging story not only highlights the complex relationship between automation and human work throughout history, but also raises doubts about whether AI can truly transform the world as threatened. The tech-heavy Nasdaq index, which had been driven almost entirely by the rise of AI-related stocks, has fallen about 8% since its peak in early June.
One critique of the “AI-will-do-everything” narrative is what we might call the O-ring argument. It comes from the explosion of the space shuttle Challenger 73 seconds into its flight on January 28, 1986. A lengthy investigation found that the multi-billion-dollar spacecraft was brought down by a rubber O-ring that failed in cold temperatures.
That cheap O-ring turned out to be critical. The analogy suggests that as long as AI can’t perform every task perfectly, it will increase the value of the tasks it can’t do. Depending on which tasks AI takes over, it could either boost the value of high-skilled workers by handling their lower-level tasks, or create more opportunities for lower-skilled workers by taking over more expert tasks.
As one recent study noted: “Despite strong substitution at the task level, overall employment effects are modest, because reduced demand in exposed jobs is offset by productivity-driven increases in labor demand at companies adopting AI.”
Things could change. As Jed Kolko points out, research on AI’s impact on the job market is still in its early stages. There are nearly four years left in Amodei’s one-to-five-year window. And maybe the real disruption comes in year six. According to the Federal Reserve, AI adoption is growing quickly across businesses.
Moreover, Autor argues, AI is getting better. And there’s no sign it will hit a ceiling anytime soon. “Skepticism about the stochastic parrot is behind us,” he told me. The dystopian AI future—or utopian, if you’re on the winning side—Owning and running the AI is still a possibility.
“Insiders are as enthusiastic as ever,” said Daron Acemoglu, the Nobel Prize-winning economist. “They still believe artificial general intelligence is just around the corner.” Indeed, Elon Musk hasn’t given up on the idea that “AI plus robots will be able to do everything, leading to universal high income. Work will be optional.”
You might recall the remark by Nobel Prize-winning economist Robert Solow during the early days of the computer revolution: “You can see the computer age everywhere except in the productivity statistics.” It took about another ten years for businesses to reorganize around the new technology. But computers did eventually show up in the data.
Still, clouds are gathering on the AI horizon. It’s not just that AI might not eliminate all human work. AI may also fail to deliver on its promise of huge economic opportunity at a price humanity is willing to pay.
The political mood has clearly turned against it. Seven out of ten Americans oppose building AI data centers in their area. While this is partly due to their huge energy demands, which drive up local electricity costs, AI’s unpopularity is also linked to the idea that it will destroy society as we know it.
There are other obstacles too. Despite major progress, there are still big doubts about whether AI can do everything a modern economy needs. “Not everything is a computational problem,” notes Autor. AI is good at copying language, but it can’t connect language to the real world around it. Despite its advances, it still makes plenty of critical mistakes.
Gig workers are endlessly exploited. AI could make more of us share their fate.
And then there’s the impossible economics. Even if AI could eventually solve all our problems, the solution looks expensive. How much of our GDP are we willing to invest in AI data centers—20%, 30%, 40%? According to some estimates, that’s where we’re headed. The International Energy Agency predicts that power demand from data centers will more than double by 2030, reaching about 945 terawatt-hours—more than Japan’s total energy consumption.
The economics look even shakier when you consider how fast AI investments lose value, as new models overtake those developed just a few months ago. Companies developing AI models “are never going to make money,” Acemoglu said. “They are losing hundreds of billions of dollars every year.”
You might dismiss Altman’s new modesty as a PR move. Maybe someone told him that linking the AI revolution to mass unemployment wasn’t smart politics. But doubts about AI’s much-touted abilities go beyond marketing. The grand, historic promise may be in trouble. Perhaps artificial intelligence just can’t deliver at a price society is willing to pay.
Frequently Asked Questions
Here is a list of FAQs about the idea that AI will cause mass job loss soon written in a natural tone with clear direct answers
BeginnerLevel Questions
1 I keep hearing that AI is going to take all our jobs Is that really true
Answer Probably not anytime soon While AI is changing how we work its much better at automating specific tasks than entire jobs Most roles involve a mix of skillslike judgment creativity and human interactionthat AI cant easily replicate
2 So is AI just a fad that wont matter for my career
Answer No its not a fad but its also not an apocalypse AI is a powerful tool that will likely become a standard part of many jobslike how computers or spreadsheets did The key is learning to work with it not worry about being replaced by it
3 What kinds of jobs are most at risk from AI
Answer Jobs that involve repetitive routine taskslike data entry basic translation or simple customer service queriesare most exposed But even those jobs often involve unpredictable situations that require a human touch
4 Should I be worried about my job if I work in a creative field
Answer Less than you might think AI can generate images or text but it lacks true creativity taste and understanding of human emotion Its more likely to become a tool that helps you brainstorm or handle boring parts of your work freeing you up for the bigpicture ideas
5 Whats the biggest reason AI wont cause a sudden jobs apocalypse
Answer The biggest reason is that AI is expensive to build and maintain and it makes mistakes Companies cant just flip a switch They need time to test it deal with legal risks and retrain staff This slow rollout gives people time to adapt
Advanced Questions
6 The article says probably not anytime soon What could actually trigger a faster job apocalypse
