Artificial intelligence is accelerating faster than society can adjust, and researchers at Anthropic are warning that this could unleash a turbulent decade—particularly for white-collar workers whose jobs are increasingly at risk of being automated.
The AI firm, which recently released some of the most advanced large language models to date, said current technology is already capable of replacing many entry-level office roles and that the situation could worsen significantly over the next few years.
Sholto Douglas, one of the company’s top researchers, told AI commentator Dwarkesh Patel that a drop in white-collar employment is not only possible—it’s likely inevitable. He said this shift could begin within two years and become impossible to ignore within five.
“There is this whole spectrum of crazy futures,” Douglas explained. “But the one that I feel we’re almost guaranteed to get—this is a strong statement to make—is one where, at the very least, you get a drop in white-collar workers at some point in the next five years.”
He added that even if AI advancement were to slow, existing models trained on domain-specific data are already capable of automating a large share of administrative, customer service, and analytical functions.
“The current suite of algorithms is sufficient to automate white-collar work provided you have enough of the right kinds of data,” he added.
Anthropic’s Dario Amodei had previously echoed similar concerns. In a recent CNN interview with Anderson Cooper, the company’s CEO warned that up to half of all entry-level white-collar jobs may disappear within five years due to AI automation. These roles include positions in customer support, paralegal work, bookkeeping, and marketing—jobs once considered relatively safe from mechanization.
Trenton Bricken, another Anthropic technical staff member, added that businesses should already be preparing for the large-scale automation of desk jobs.
“We should expect to see them automated within the next five years,” he said.
Evidence of disruption is already surfacing across the corporate world. Companies like Shopify and Duolingo have begun reducing hiring for functions that AI now handles, such as content creation, basic coding, and support ticket resolution. Meanwhile, Klarna, the Swedish fintech company, briefly embraced AI to replace over 700 customer service staff but had to walk back some of that adoption due to performance and trust issues.
Revelio Labs, a workforce analytics firm, reported a steep drop in job postings for high-exposure white-collar roles such as data entry, IT support, and financial analysis—positions that generative AI models can now increasingly perform with minimal human supervision. Firms seeking to cut costs have already begun turning to tools like OpenAI’s GPT-4, Google’s Gemini, and Anthropic’s Claude series to automate documentation, data processing, and even aspects of software engineering.
Nvidia CEO Jensen Huang recently added his voice to the growing chorus warning about job disruptions. Huang, whose company produces the GPUs powering most AI systems, noted that “every job” will be touched by AI. He predicted that while some employees may not be outright replaced, they could still find themselves edged out by more productive colleagues leveraging AI tools to outperform human-only workflows.
Anthropic’s new models, Claude Opus 4 and Claude Sonnet 4 are already showing signs of this future. Early testing shows these systems can write software code, produce legal drafts, summarize long reports, and even troubleshoot technical problems with minimal input. One tester said Opus 4 “coded autonomously for nearly seven hours” on a complex task, offering a glimpse into how AI could displace even highly skilled knowledge workers.
The looming wave of automation has raised deeper societal concerns. Unlike previous waves of disruption that primarily affected factory or field workers, this one is targeting the middle class—those with college degrees and office careers. Economists say this shift could spark mass underemployment and wage stagnation, especially in countries where the service sector dominates.
At the same time, the cost savings AI offers are hard for companies to ignore. Replacing a team of entry-level analysts with a single AI model fine-tuned on internal data can slash operational expenses, boost productivity, and streamline decision-making. But the social fallout of mass job losses may force governments to intervene.
Calls for stronger labor protections, AI taxation, and universal basic income have resurfaced as policymakers grapple with how to manage the transition. Some experts are urging the creation of retraining programs and education reforms aimed at equipping workers with skills in robotics, machine learning, and human-centered design—roles still out of reach for today’s AI.
The timeline laid out by researchers like Douglas and Amodei is already ticking. As they warn, the transition could be brutal before it gets better. If robotics, which lags behind software AI, fails to keep pace, displaced workers may find themselves unable to pivot into new roles that demand physical labor or technical expertise. In that gap, the world may experience what Douglas called a “pretty terrible decade”—a painful reality where jobs vanish faster than new opportunities emerge.
“Imagine a world where people have lost their jobs, and you haven’t yet got novel biological research. That means people’s quality of life isn’t dramatically better,” he said. “A decade or two after, the world is fantastic. Robotics is solved, and you get to radical abundance.”
