Tech CEO rings alarm bells, predicts AI to replace entry-level jobs within few years

The CEO of one of the world’s leading artificial intelligence labs is raising alarm about the rapid advancement of AI and its potential to trigger a sharp rise in unemployment—sooner rather than later.

Dario Amodei, CEO of Anthropic, cautions that both policymakers and corporate leaders are unprepared for the profound labour disruptions ahead. “AI is starting to get better than humans at almost all intellectual tasks, and we’re going to collectively, as a society, grapple with it,” Amodei told CNN’s Anderson Cooper in an interview. “AI is going to get better at what everyone does, including what I do, including what other CEOs do.”

In a separate conversation with news outlet Axios, he warned that within the next one to five years, AI could eliminate up to half of all entry-level white-collar jobs, potentially driving U.S. unemployment to 20%. This would represent a fivefold increase from current levels, nearing the record highs briefly seen during the peak of the Covid-19 crisis. Warning voices by industry grow louder Warnings about AI-driven job loss are not new.

Economists and academics have been forecasting disruptions to employment due to automation for years. A World Economic Forum report from earlier this year revealed that 41% of surveyed employers expect to reduce their workforce by 2030 because of AI. However, Amodei’s projections stand out, both because of his prominent role in the industry and the scale of the disruption he envisions. Anthropic, like other AI companies, is developing technology capable of working hours on end with minimal human intervention.

The company's newest AI model, for instance, can operate independently for nearly seven hours, handling increasingly complex tasks. Amodei believes this shift will impact not just lower-skilled roles but also white-collar positions that traditionally required significant education and training. The traditional view of technological progress assumes that automation targets routine or lower-skilled jobs, while displaced workers move into more specialized, better-paying roles. But Amodei suggests AI may upend that model. “If AI creates huge total wealth, a lot of that will, by default, go to the AI companies and less to ordinary people,” he said. “So, you know, it’s definitely not in my economic interest to say that, but I think this is something we should consider and I think it shouldn’t be a partisan thing.”

He hinted that taxing AI companies could be one potential solution. Some researchers believe the impact will be broad, affecting professionals such as paralegals, payroll clerks, financial advisers, and software developers. Tech leaders are already acknowledging these shifts. Meta CEO Mark Zuckerberg recently stated that AI may soon write half of the company’s code, while Microsoft CEO Satya Nadella said AI is currently generating about 30% of theirs. According to Amodei, Anthropic tracks how users engage with its tools.

Currently, 60% of people are using AI to augment their jobs, while 40% are automating them outright—a figure that is growing. This trend illustrates the creeping shift from collaboration to replacement. Despite these concerns, Amodei encourages the public to embrace AI rather than ignore it. “People have adapted to past technological changes,” he said. “But everyone I’ve talked to has said this technological change looks different, it looks faster, it looks harder to adapt to, it’s broader. The pace of progress keeps catching people off guard.” He advises “ordinary citizens” to “learn to use AI.” 

While predictions about AI’s speed of development differ, some experts argue that the pace may slow as companies exhaust high-quality training data available on the public internet. Others contend AI will likely automate tasks rather than entire jobs, allowing humans to focus on what machines still struggle with. Nevertheless, there is growing consensus among economists and technologists that the time to prepare for AI’s economic impact is now—regardless of whether the disruption arrives in one year or ten. 

By Nazrin Sadigova