AI developer Anthropic has launched the Anthropic Economic Index, featuring first-of-its-kind data and analysis based on real-world conversations with its AI model, Claude, providing the clearest picture yet of how AI is being incorporated into the modern economy.

According to the Index’s first report, based on one million anonymised user inputs, software development and technical writing are occupying the lion’s share of tasks, well above any other activity, but unsurprising, given that 37.2% of all queries sent to Claude came from those working in computing and mathematics.

Computer programmers are so far using AI the most (6.1%), followed by systems software developers (5.3%), and application devs (3.4%), with the top queries sent to Claude covering tasks like software development and website maintenance (16.8%), code debugging (6.9%), and the designing of new database systems (2.3%).

Although AI is right now mostly being used to automate repetitive tasks and provide a technical boost, Anthopic’s data also shows a growing trend of genAI use by those in the creative industries.

Queries covering the arts, design, sports, entertainment, and media represent the second largest category of AI use (10.3%), with technical writers, copy writers and editors using AI the most, for tasks including PR management (1.3%) and the creation of marketing and promotional strategies (1.2%).

At the other end of the scale, AI use was unexpectedly low among those working in business and financial occupations. Just 0.8% of users asked Claude to help analyse financial data or develop investment strategies, while even fewer (0.4%) turned to AI to record, analyse and report operational or research data.

Of course, this does not mean that AI is having less of an impact on business than previously thought, as many firms are taking pains to equip themselves with tailored solutions, rather than rely on publicly available models like those offered by Anthropic.

However, the Index does show that very few occupations use AI in much depth across their tasks, with only 4% of jobs using AI for at least 75% of their tasks. According to Anthropic’s data, more moderate use of AI is much more widespread, with 36% using it for at least 25% of their work.

That could alleviate some concerns about the prospect of AI replacing human workers, with Anthropic finding no evidence that any jobs have so far been entirely automated, with AI instead being diffused across many occupations across the economy.

In fact, more than half (57%) of all tasks were augmented by AI, where the technology collaborates with a user, versus 43% where the model directly performed the task itself, with users mostly engaging Claude to help with learning and validation, such as double checking documents they’ve already written.

According to the researchers, using AI models in this way is evidence that most people are employing the tech as a productivity tool, which could inform the development of more collaborative AI interfaces that promote meaningful engagement.

Recommended reading

• Could DeepSeek Topple the US’s AI Market Dominance?
• Is the UK Lagging Behind in Global AI Race?
• UK Gov Partners with Anthropic, Focuses on AI Security

“While these patterns are informative, they capture just the beginning of AI’s integration into work,” concludes Anthropic’s report.

“As AI systems expand beyond text to handle video, speech, and physical actions through robotics, and as AI agents become more capable of carrying out extended tasks autonomously, the nature of human-AI collaboration is poised to transform dramatically.

“The challenge ahead lies not just in measuring these changes, but in using our understanding of them to help shape a better future.”