AI collaboration not automation dominates workplace use, studies find

Two studies released Thursday point to AI collaboration not automation as the main pattern emerging in workplaces, with the technology spread across many occupations but used for a narrower share of day-to-day work. Google said its first AI & Economy ATLAS found AI activity in 68% of occupations, while the tool was used for about 21% of tasks in a typical job.

Google said in a blog post that the study drew on 15 million aggregated and de-identified human-AI interactions across its AI products and tools. The company said the analysis was designed to show how people are using the technology in practice, rather than measuring exposure through job descriptions alone.

The findings suggest that workers are using AI most often as a support tool. Google cited uses including idea generation, strategy, information retrieval and learning. Fewer than 10% of AI work interactions fully automated a task, according to the company.

Google’s report said only 3% of occupations showed AI use across more than 75% of their tasks. The occupations in that group included software quality assurance analysts and testers, human resources specialists and document management specialists, according to the report.

How are companies using AI at work?

Companies are mainly applying AI to parts of jobs, rather than replacing whole roles, according to the studies. In practical terms, that means an employee may use an AI system to draft, search, compare, summarize or test work, while people still decide what to ask, how to judge the answer and how to apply it.

A separate study commissioned by Adecco Group, the Switzerland-based talent and technology company, reached a similar conclusion. Adecco said in a release that AI is altering tasks and skill requirements more quickly than it is removing jobs.

Adecco said that three and a half years after ChatGPT became available, employment rates remain at record highs across 38 member countries of the Organization for Economic Co-operation and Development. The company also said employment in AI-exposed jobs has remained stable and that 1.9 million new AI-related jobs have been created.

The adoption picture remains uneven. Adecco said fewer than 10% of U.S. firms have integrated AI into their core workflows at scale. Where companies have deployed the technology, the firm said it is changing workflows and the skills required for roles faster than it is eliminating occupations.

Denis Machuel, Adecco Group’s chief executive, said in the release: “The next phase of AI will be won by redesigning work — understanding what people should do, what AI agents can do, how skills are evolving and what skills will be needed in the future, and how value is created and measured.”

The findings add to a broader debate over how automation will reach lower-paid and frontline work. A PYMNTS Intelligence report, “Wage to Wallet Index: The Resilience Deficit: Labor Workers in an Automated Economy,” found that AI and automation are moving beyond higher-paying roles and into the labor economy.

Taken together, the studies describe a labor market in which AI adoption is visible but incomplete. The immediate effect reported by Google and Adecco is concentrated at the task level, with companies still working out how to redesign jobs, measure output and train employees as the technology spreads.

This story draws on original reporting from PYMNTS.