When your own engineers start worrying that your product might make them obsolete, you've either built something genuinely transformative or created a very awkward workplace dynamic. For Anthropic, it might be both.

The Productivity Paradox

Anthropic released findings Tuesday from an internal study conducted in August examining how employees use Claude Code, its AI coding assistant. The company surveyed 132 engineers and researchers and conducted 53 interviews to understand how the tool was reshaping daily work.

The results paint a complicated picture. On one hand, employees felt more capable across the full technology stack and significantly more productive. Twenty-seven percent of tasks completed with Claude's help were projects that simply wouldn't have happened otherwise—think scalable builds or data dashboards that seemed too time-consuming without AI backup.

Workers reported they could "fully delegate" up to 20% of their workload to Claude, especially the repetitive or tedious tasks that were straightforward to verify. That's the dream, right? Let the robots handle the boring stuff while humans focus on the interesting problems.

The Human Cost

Except here's where it gets uncomfortable. "Some found that more AI collaboration meant they collaborated less with colleagues," the study noted. One employee put it plainly: "I like working with people, and it's sad that I need them less now."

That's not just about workplace friendships. Several engineers worried their technical skills could atrophy. "When producing output is so easy and fast, it gets harder and harder to actually take the time to learn something," one engineer explained.

The existential dread was even more pointed. "I feel optimistic in the short term, but in the long term I think AI will end up doing everything and make me and many others irrelevant," one employee told researchers.

The Bigger Automation Debate

Anthropic's findings land in the middle of an ongoing debate among tech leaders about AI's impact on work. In September, OpenAI CEO Sam Altman predicted artificial intelligence would replace up to 40% of work tasks in the near future. He framed it as reshaping tasks rather than eliminating entire jobs, with some roles transforming, new ones emerging, and others disappearing.

Kevin O'Leary took a more optimistic stance that same month, dismissing fears of widespread job loss. He argued AI creates better opportunities by removing repetitive work, noting that more than 50 of his companies used AI to cut costs and boost productivity. He compared the shift to past technological transitions like television and radio.

In August, ServiceNow Inc. (NOW) CEO Bill McDermott said AI had already begun transforming corporate operations, with AI agents managing IT support, customer inquiries, and security tasks around the clock. He said the company slowed hiring for what he called "soul-crushing jobs" and predicted more companies would reorganize around AI-driven workflows.

What makes Anthropic's study particularly striking is that it's coming from inside the house. These aren't outside observers speculating about AI's impact—they're the people building the technology, watching it change their own work in real time, and they're genuinely uncertain about what it means for their futures.