Google Research has released findings from a study analyzing how workers are using its Gemini AI model. The study, called 'AI & Economy ATLAS,' examined 15 million anonymized AI interactions across Gemini App, Google's AI Mode, and the Gemini API. The researchers found that while AI is being used across a wide range of occupations, its integration into daily work remains limited and collaborative. According to the study, only 21% of all work-related tasks were classified as 'Gemini tasks'—those that met a minimum threshold of 25 related interactions. This suggests that AI is currently serving as a complement to existing work rather than a replacement for human labor. Source: arstechnica
The study also looked at how deeply AI was being integrated into different jobs, measuring the frequency of specific work tasks attempted using Gemini. Across the entire database, only 21% of all work-related tasks were classified as 'Gemini tasks,' meeting a minimum threshold of 25 related interactions. For many occupations, less than 25% of O*NET tasks saw significant attempted assistance from Gemini, indicating that humans are still primarily responsible for the majority of job components. Only 3% of occupations had Gemini being regularly consulted for at least three-quarters of the job's relevant tasks, with roles like software quality assurance analysts and testers falling into this category. Source: arstechnica
The researchers also examined the types of tasks Gemini users were asking the model to perform. Cognitive tasks, such as drafting and generating ideas or retrieving information, made up 86% of Gemini interactions by volume. However, these tasks were typically low-expertise and did not require a high level of expertise. The study found that workers were more likely to use Gemini for routine cognitive tasks rather than complex, high-expertise work. This suggests that AI is currently being used to augment human work rather than replace it. Source: arstechnica
