Speaking to AI podcaster Dwarkesh Patel, Anthropic’s Sholto Douglas said he predicted there would be a “drop in white-collar workers” over the next two to five years, even if current AI progress stalls.

“There is this whole spectrum of crazy futures. 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 said. “I think it’s very likely in two, but it seems almost overdetermined in five.”

“The current suite of algorithms is sufficient to automate white-collar work provided you have enough of the right kinds of data,” he added.

Trenton Bricken, a member of the technical staff at Anthropic, seconded his fellow researcher’s point, saying: “We should expect to see them automated within the next five years.”

The discourse around AI job losses has been heating up recently, with some major tech figures acknowledging that the technology will have at least some effect on desk jobs.

In an interview with CNN’s Anderson Cooper last month, Anthropic CEO Dario Amodei predicted that within five years, AI could automate away up to 50% of all entry-level white-collar jobs.

Nvidia’s Jensen Huang has also said “every job” will be affected by AI, but predicted that workers would be more likely to lose their jobs to an AI-enhanced colleague rather than have it purely automated.

Companies like Shopify and Duolingo are already slashing hiring for roles AI can handle.

According to Revelio Labs data cited by Business Insider, there has also been a steep drop in job postings for high-exposure positions like IT and data analysis.

While some companies, like fintech Klarna, have walked back aggressive AI adoption due to quality concerns, most seem to be committed to using some form of AI to shrink white-collar workforces.

AI is already automating some work

AI is already proving it can handle coding and a wide range of desk jobs, raising the possibility of a future where machines do the thinking, and humans are left with the hands-on work.

Douglas said this scenario could lead to a “pretty terrible decade” before things start to improve for the better.

“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.”

Anthropic has recently unveiled its latest generation of cutting-edge AI models, Claude Opus 4 and Claude Sonnet 4.

The models represent a significant leap in AI’s coding ability, beating out Google and OpenAI’s most advanced offerings.

One early tester of Claude Opus 4 said the model “coded autonomously for nearly seven hours” after being deployed on a complex project.