Workers are using artificial intelligence more for augmentation (57%) than automation (43%), according to an analysis of over 4 million AI user prompts by AI safety and research company Anthropic. The study, part of the newly launched Anthropic Economic Index, aims to track AI’s long-term impact on labor markets and the economy.
AI Adoption by Occupation
Computer and mathematical workers, particularly software engineers, accounted for most AI usage. This group made up 37.2% of AI-related requests, with common tasks including software modification, code debugging, and network troubleshooting.
Despite concerns about AI replacing jobs, Anthropic found that only 4% of occupations used AI for at least 75% of tasks, while about a third of jobs used AI for at least 25% of tasks. The company stated, “There wasn’t evidence in this dataset of jobs being entirely automated.” Instead, AI is dispersed across various tasks, impacting some more than others.
Who Uses AI the Most?
AI use was highest among workers in mid-to-high salary ranges, such as computer programmers and copywriters. Low-wage and very high-paying jobs had low AI adoption, especially those requiring significant manual dexterity, such as shampooers and obstetricians.
Among other top AI users:
- Bioinformatics technicians (2.9%)
- Technical writers (1.8%)
- Copywriters (1.6%)
- Tutors (1.6%)
- Archivists (1.5%)
How AI Is Used: Augmentation vs. Automation
Even in cases of heavy AI usage, it was primarily used as an augmentative tool rather than replacing human skills. Workers leveraged AI for:
- Collaboration – Checking work, brainstorming, and learning new skills
- Automation – Completing tasks with minimal interaction or environmental feedback
Anthropic noted that usage patterns could be ambiguous. For instance, a user may request a fully generated memo (classified as automation) but later refine it manually (augmenting their work).
Future Outlook
The findings align with prior assessments, such as Indeed’s 2023 report, which found no skills AI was “very likely” to replace. While AI’s workplace impact is evolving, Anthropic’s early data suggests that most current jobs will adapt rather than disappear, with AI acting as a productivity tool rather than a replacement.
