Only 4% of workers in the occupation use AI in three-quarters of tasks, according to human studies exploring how they use it.
The survey found that approximately 36% of occupations incorporate AI into at least 25% of tasks. These findings are consistent with previous reports showing that few companies have adopted the technology fully.
Of those seeking AI support, approximately 37% in the role of software engineering, 10% in the field of media, arts and design, and 9% are involved in education and library services.
The occupational roles in which AI is least useful tend to involve physical labor, such as transportation and material mobility, medical assistance, agriculture, fishing and forestry occupations.
The study found that 57% of AI are directed at enhancing human work, while 43% automate work.
To understand how AI is affecting the economy, humanity, one of the top candidates in the commercial AI space behind Openai, has launched the Human Economic Index.
Supported by research papers [PDF] On this subject, the initiative focuses on the impact AI has on certain occupations. This is defined by the U.S. Bureau of Labor Statistics.
According to costumes, AI has influenced people's work, and is an issue that has prompted many research in recent years, raising troubling socio-political questions about labor costs, wages and the outcomes of automation.
When France holds an AI Action Summit to discuss these issues, Anthropic provides data on actual AI usage based on prompts received from users of software with appropriate privacy protection.
To date, the impact of AI has not been particularly widespread, but it has changed certain jobs.
“Only about 4% of occupations show AI use in at least 75% of tasks, suggesting the possibility of deep task-level use in several roles,” the paper cites teachers in foreign languages and literature as examples. “More broadly, it shows that around 36% of occupations are used in at least 25% of tasks, indicating that AI is already beginning to spread into the task portfolio in a significant portion of the workforce.”
Computer and mathematics occupations were the most cited, but those occupations account for only about 3.4% of US workers. Category Office and Management Support was responsible for just 7.9% of Claude prompts, while representing the total workforce of 12.2%, the highest percentage of US workers.
AI use peaks in moderate to high-paying occupations, particularly IT-related jobs. But they fall out at both extremes of the salary spectrum, such as high-paying roles like doctors and low-paying jobs like restaurant workers.
“Our empirical findings examine and challenge previous predictions regarding the impact of AI on work,” says Anthropic's paper. “Web [2019, PDF] We predicted the highest AI exposure in occupations around the 90th wage percentile, but peak usage was observed in moderate-to-high wage occupations, with significantly lower usage at both poles of wage distribution. This pattern suggests that factors beyond technical feasibility, such as implementation costs, regulatory barriers, and organizational preparation, may be driving adoption in the highest wage sector. ”
Human researchers also point to predictions from a 2023 research paper that suggests that 80% of US workers may have at least 10% of work tasks affected by the introduction of LLMS. The AI giants show that their data shows that around 57% of occupations use AI for at least 10% of tasks, but suggests that numbers could increase as AI adoption grows. ®
