Claudeユーザー調査：新機能が速度を上回りAIの最大の恩恵に、クリエイターは取り残され感を抱く

A survey of 81,000 Claude users shows that gaining new capabilities ranks slightly ahead of speed as the most common productivity benefit. Creatives, meanwhile, feel both limited and threatened by AI. The sample, however, has a significant bias.

When AI changes work, it's not primarily through speed. That's what a survey by Anthropic of 81,000 Claude users suggests. Among those who described specific productivity effects, 48 percent named an expansion of their skill set as the biggest benefit, while 40 percent pointed to pure speed gains. The gap isn't huge, but the fact that scope ranks ahead of speed at all is notable: AI appears to unlock entirely new capabilities slightly more often than it makes existing tasks faster.

The sample is far from representative, though, which the authors themselves acknowledge. Only users with personal Claude.ai accounts who volunteered to participate were surveyed. People using AI on their own or as solopreneurs are more likely to report new capabilities than someone whose employer handed them the tool to get existing work done faster. Enterprise users are completely absent from the data. That likely skews both the high weight given to scope over speed and the finding that productivity gains mostly benefit the users themselves.

The examples in the study illustrate this bias indirectly: a delivery driver uses Claude to start an e-commerce business, a landscaper builds a music app. These are classic side projects from intrinsically motivated individual users. The study also doesn't measure how successful these projects actually are. When one respondent without a technical background describes themselves as not a "techie" but now a "full-stack developer", some skepticism is warranted.

Still, for this group, AI functions as a gateway to work that was previously out of reach, whatever the quality of the output.

Highest and lowest earners report the biggest gains

The distribution of productivity gains by income is surprising, according to authors Maxim Massenkoff and Saffron Huang. Both the highest-paid and lowest-paid occupational groups report the largest benefits. Management roles come out on top, followed closely by computer and math occupations. The result for high earners holds even when IT jobs are excluded, the study says.

The fact that low-wage workers also report high gains fits the bias described above: many of them aren't using AI for their actual jobs but for technical side projects. The delivery driver with the e-commerce business shows up in the statistics alongside the software developer, even though the two are using AI in different contexts.

Creatives show the opposite dynamic. Visual artists and writers find AI too rigid and limiting to help with their own work, according to the study. At the same time, they worry that the spread of AI in creative fields will hurt their job prospects. This finding is reflected in a U-shaped relationship between perceived speedup and job anxiety: even those who say AI slows them down express above-average concern about being replaced. On the other end, anxiety also rises among those experiencing the biggest speed gains. When you see how quickly AI handles your tasks, you start wondering how long the job will last.

Overall, one in five respondents expressed concern about job loss, with early-career workers doing so far more frequently than experienced professionals. The average productivity rating was 5.1 out of 7. Most respondents also said the benefits went to them personally rather than their employer which isn't surprising given the complete absence of enterprise users in the sample.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.

Subscribe now