AI is mainly automating work instead of enhancing it, which is leading the technology to be a catalyst for replacing jobs, according to a new study.

An AI startup, Anthropic, recently valued at $183 billion, unveiled a report indicating that over 77% of businesses utilizing their tool, Claude, do so for task automation. In contrast, only 12% used it to enhance or collaborate on tasks.

“The 77% figure highlights that companies prefer using Claude to offload tasks, rather than for teamwork purposes,” the report highlighted. “This prevalent trend in automation might lead to significant shifts in the labor market, potentially impacting workers whose roles are susceptible to being automated.”

According to the report, businesses predominantly deploy Claude for coding and administrative duties. Claude generates code effectively, similar to platforms like Replit and Cursor, that convert text prompts into developers’ code. These capabilities could potentially assume the coding responsibilities of software engineers. At a Council on Foreign Relations event in March, Anthropic CEO Dario Amodei foretold that within a year, AI could entirely code for software engineers.

“In 12 months, we may be in a world where AI is writing essentially all of the code,” Amodei said at the event.

^{Anthropic CEO Dario Amodei. Photo by Chance Yeh/Getty Images for HubSpot}

The report further stresses the risk of mass unemployment and job displacement due to automation by AI. Amodei commented earlier this year, foreseeing in May that AI could potentially eliminate 50% of entry-level, white-collar positions within the next five years, leading to 10-20% unemployment rates, affecting sectors such as law, technology, and finance.

Anthropic’s Head of Economics, Peter McCrory, mentioned to Bloomberg that it remains unclear if the trend towards automation arises from “new model capabilities” allowing AI to assume more responsibilities or if it stems from increasing “comfort levels” with AI, leading more tasks to be delegated to Claude.

Essentially, researchers are uncertain if the surge in automation is due to AI’s advancing abilities or a growing willingness among users to embrace the technology.

Understanding the reason presents “an important area of research for the future,” McCrory told the outlet.