A new report by artificial intelligence firm Anthropic suggests that AI is currently assisting workers rather than replacing jobs, challenging fears that technology could eliminate large numbers of roles. The study analyzed how people and companies used Anthropic’s AI assistant, Claude, drawing on an anonymized sample of two million conversations from November 2025.
The research looked beyond the frequency of AI use, focusing on the types of tasks assigned and the success rate of AI in completing them. To measure impact, authors introduced “economic primitives,” which evaluate the difficulty of tasks, the level of education required to understand the user request and AI response, the autonomy granted to the AI, and how reliably it completes assignments. According to the report, these measures provide a “new window for understanding AI’s impact on the economy.”
Findings show that 49 percent of jobs can now use AI for at least a quarter of the tasks involved, a 13 percent increase from early 2025. However, the effects vary across roles. While AI can upskill workers in some occupations, in others it removes the most skill-intensive tasks, and in some cases the least, creating a complex reshaping of work rather than wholesale job loss.
AI usage remains concentrated in specific areas, particularly coding and software development. Tasks supported by AI generally require higher levels of education than the average in each economy. Geographic differences also emerged. Higher-income countries tend to use AI more frequently for work and personal tasks, while lower-income countries show a higher share of educational use. Anthropic noted this reflects different stages of adoption, with users in wealthier nations expanding AI usage into everyday life beyond work.
The study distinguished between tasks automated entirely by AI and those augmented through collaboration. About 52 percent of work-related conversations involved augmented tasks, such as co-writing documents, though this share has fallen slightly since January 2025. Complex tasks, however, were less reliably completed by Claude, with success rates declining as task difficulty increased. This reduces the time humans ultimately save and highlights the need for human oversight.
The report is the fourth edition of Anthropic’s economic index, which tracks AI integration and its effects on productivity and jobs. Authors emphasized that understanding how AI is used is as important as measuring its adoption. They noted that experimentation by users, along with regulatory frameworks that balance safety and innovation, will shape how AI transforms economies.
“Claude struggles on more complex tasks: As the time it would take a human to do the task increases, Claude’s success rate falls,” the authors wrote, underscoring the limits of current AI capabilities.
Overall, the study paints a nuanced picture: AI is helping workers and reshaping tasks, but its potential to replace jobs is uneven and depends on task complexity, occupation, and local adoption patterns.
