Half of Claude Users Say AI Can Replace Half Their Work

According to a survey conducted by Anthropic of approximately 9,700 Claude users, roughly half of respondents answered that "AI can currently handle 50% or more of their work." This serves as one indicator of how deeply AI has become integrated into daily tasks.

As the use of AI assistants expands, how users themselves evaluate AI's ability to replace work has become a topic directly linked to enterprise and individual AI adoption decisions. Anthropic's publication of a large-scale user survey appears to be driven by growing interest in this area. While this survey is based on users' actual experience and differs from objective productivity measurement, it holds certain significance in understanding perceptions in the field.

Looking at the survey details, 26% of respondents predicted that AI would cover 60-90% of their work within the next 12 months. In other words, more than one in four people hold the outlook that AI will handle the majority of work within a year. Conversely, a defining characteristic of this survey is the coexistence of both optimistic views and concerns.

When examining respondents by attribute, it became clear that younger workers with less career experience tend to feel stronger anxiety about AI. In contrast, "heavy users" who use AI more frequently demonstrated the most optimistic stance toward their career and future prospects. This gap suggests that the more actual contact one has with AI, the easier it becomes to perceive it as a "tool" rather than a threat.

The implications of these survey results are significant for companies and working people. If the perception that AI can handle half of work tasks becomes widespread, it may impact workplace AI adoption policies, personnel planning, and approaches to skills development. Particularly, addressing the concerns of younger employees is emerging as an important organizational challenge.

However, it is important to note that this survey is based solely on users' subjective evaluations and does not directly measure actual impact on work. How the sense that "AI can accomplish half the work" translates to improvements in productivity or changes in employment structures requires future research and empirical data. It is important to reference the survey results as a field perspective while avoiding overly simplistic interpretations.

Alongside the evolution of AI technology, users' perceptions and expectations regarding AI are changing rapidly. Moving forward, attention should focus on how these subjective evaluations align with actual changes in working practices and employment structures, and how to address concerns centered on younger workers. As development companies like Anthropic continue to conduct user surveys, the actual state of AI utilization is expected to become increasingly clear.