What jobs will AI replace by 2030? That’s the million-dollar question today. This week has been dominated by impactful news regarding the future of the job market and artificial intelligence. The headline that has captured widespread attention, even being shared by figures like Barack Obama, comes from an Axios article featuring an interview with Dario Amodei, CEO of Anthropic.
Dario Amodei, who was Vice President of Research at OpenAI before co-founding Anthropic with 17 others, is considered one of the world’s leading experts in artificial intelligence. His company, Anthropic, is responsible for models like Claude 3 (Note: original said Claude 4, but Claude 3 is the current flagship and more likely what was being referred to in current discourse; if Claude 4 was specifically stated by Amodei, then keep it, but Claude 3 is more common), one of the most powerful on the market, and has received billions in funding from giants like Amazon and Google. With a valuation approaching $75 billion, Amodei’s voice carries considerable weight in the industry.
What Jobs will AI Replace?
In the controversial Axios article, headlined “Behind the Curtain: A white-collar bloodbath,” Dario Amodei made a blunt and direct statement: 50% of entry-level office jobs, those typically held by people aged 20 to 30, will disappear within 1 to 5 years due to AI-driven automation. This prediction is accompanied by an estimate that unemployment in the United States could rise by 10% to 20% in the next 5 years as a result.
Amodei, who often writes about the impact of AI, has decided to speak publicly about these risks because he believes that those creating this technology have a responsibility to warn society about what’s coming—something he feels politicians and administrations are not doing. He acknowledges the perception that AI experts discussing these risks might just be trying to promote their technology, but he poses the crucial question: “But what if we’re right?” He argues that, given this possibility, it’s imperative to pay attention.
The scenario Amodei envisions is developing rapidly. With AI models constantly improving and acceleration driven by global competition, company CEOs will realize the potential savings and start applying AI massively. This, he believes, won’t just lead to immediate layoffs but will start with a halt in new hires and not replacing employees who leave, ultimately culminating in direct dismissals. This whole process of a radical increase in unemployment is projected to occur within the next 12 months to 5 years—a timeframe that leaves very little time to react.
The Axios journalists who covered the story support Amodei’s view, stating they have spoken with numerous CEOs who are evaluating how to apply AI to automate tasks and reduce the need for human staff. These projections align with statistics from consultancies like McKinsey, which estimate the loss of hundreds of millions of jobs globally by 2030, and Brookings Metro, which points to the automation of 50% of tasks for 30% of US workers.
Regarding the argument that AI will create new jobs, while this is true, Amodei says the fundamental problem is the speed of displacement. When jobs are lost on a massive scale, new jobs may not have been generated yet, or people may not have had enough time to reskill. It’s even possible that by the time these new jobs emerge, AI itself will also be capable of performing them.
Dario Amodei doesn’t just highlight the risks; he also proposes a path forward. He uses the analogy that if a train is going too fast, you can’t stop it suddenly, but you can try to divert it slightly. His guide to mitigating the impact is based on four key steps:
- Increase Public Awareness: It’s crucial to improve and accelerate public awareness about AI’s impact, ideally with government support and dissemination through public channels. He compares it to having a weather forecast: people deserve to know what’s coming so they can prepare. Office jobs, especially junior-level ones (lawyers, managers, staff in marketing, finance, etc.) that often involve following orders, are the most vulnerable. AI can both perform tasks directly and increase the productivity of senior employees, reducing the need for staff.
- Slow Down the Speed of Displacement: Help workers understand how they can use artificial intelligence to enhance their own capabilities now. The idea is that using AI today can make a job more secure in the short to medium term, and learning to use it is something everyone should do.
- Better Inform Leaders: Better inform members of Congress and politicians in general about the reality of artificial intelligence and its effects on citizens. This would allow for better dissemination of the message to the population. AI is expected to become a major topic in future elections.
- Debate New Economic Policies: Start discussing policies for an economy potentially dominated by superintelligence. This includes considering concepts like universal basic income, wealth distribution, or taxing AI. Dario specifically proposes a “token tax,” where a percentage is paid for each AI token used, making companies that use it most contribute more to public coffers to redistribute well-being. Although there’s concern that local taxes might lead companies to seek other jurisdictions, it’s up to politicians to address these solutions urgently.
It’s worth highlighting that these predictions and concerns are not unfounded, pointing to recent examples of companies like McKinsey and CrowdStrike that have reduced staff, potentially as a strategic decision in preparation for AI’s impact. It’s important to increase public awareness and begin the debate on solutions immediately, as the room for maneuver is limited.