Google has released its inaugural AI & Economy ATLAS report, examining how generative artificial intelligence (AI) is being used in real-world work environments. Based on nearly 15 million anonymised interactions across its services, including Gemini, AI Mode and the Gemini application programming interface (API), the US technology giant examined AI adoption on a large scale.
Rather than finding widespread job automation, the report suggests AI is primarily being used to improve productivity and support decision-making. The report aims to understand how people interact with AI systems while performing their jobs.
Rather than relying on surveys or projections, Google said the analysis is based on actual usage across its AI products, including Gemini, AI Mode and the Gemini API. The findings provide a practical picture of AI adoption, focusing on how workers use the tools to complete tasks rather than how AI could theoretically be used.
While AI is often associated with office-based jobs, the report shows its use is extending to physical and manual work.
Although nearly one-third of highly physical occupations showed no observed AI usage, workers in several manual and technical trades are already using AI to support their daily work.
According to the report, AI is commonly used for diagnostics, troubleshooting and real-time learning rather than replacing hands-on tasks.
The report also highlights greater use of multimodal AI, which can process images and videos, in such occupations.
It cites automotive technicians and industrial mechanics using AI to interpret complex test results, identify electrical wiring problems and inspect machinery for wear.
The use of multimodal AI in these roles is more than twice the overall workplace average, suggesting the technology is increasingly serving as a practical support tool in technical work environments.
AI supports workers rather than replacing them
One of the report’s main conclusions is that AI is currently being used far more as an assistant than as a replacement for workers.
According to the report, fewer than 10 per cent of AI interactions involve complete task automation. Most users instead rely on AI as a copilot that helps them perform parts of their work more efficiently.
People most commonly use AI for:
- Drafting written content
- Summarising information
- Conducting research
- Coding assistance
- Brainstorming ideas
- Analysing information
- Solving problems
These are tasks in which AI speeds up a process or offers suggestions, while users continue to review the output, make changes and take final decisions.
The findings challenge the assumption that AI is already replacing large numbers of workers.
Instead, the report suggests adoption is currently centred on improving how people work rather than eliminating the need for human involvement.
AI reaches more jobs
Another key finding is that AI is being used across every major industry, from professional services and construction to hospitality.
It is also present in more than 68 per cent of occupations, covering jobs ranging from software developers and market researchers to farmers, industrial engineers and foresters.
Beyond work, AI is increasingly being used for activities such as education, household tasks, travel and leisure. AI-related conversations cover activities that account for 98 per cent of Americans’ waking hours.
Despite its growing presence across occupations, AI use within most jobs remains selective.
According to the report, AI is used for only 21 per cent of tasks in the median occupation.
Only 3 per cent of occupations use AI for more than three-fourths of their tasks. These include software quality assurance analysts and testers, human resources specialists and document management specialists.
The findings indicate that occupations are changing as AI becomes part of selected workflows rather than replacing entire roles.
Instead of transforming every aspect of a job, AI is currently being applied to specific tasks where it can save time, organise information or generate initial ideas.
The report also finds that AI adoption does not increase uniformly with expertise.
While high-income workers with specialised skills are among the most frequent users, the technology is used more often for tasks requiring low- to mid-level expertise than for those demanding the highest level of expertise.
Why training and workflow changes may matter more than job cuts
The findings point to a different organisational priority from large-scale workforce reductions.
Since AI currently supports only parts of most jobs and complete automation accounts for fewer than 10 per cent of interactions, organisations may need to focus on helping employees use the tools effectively.
AI use varies widely across occupations
Around 30 per cent of occupations use AI in at least a quarter of their tasks, while 11 per cent use it in half of their tasks.
Only 3 per cent of occupations use AI for at least three-fourths of their tasks, including software quality assurance analysts and testers, human resources specialists and document management specialists.
Organisations may, therefore, need to place greater emphasis on AI training, redesigning workflows and reskilling employees so the technology becomes part of existing work processes.
Rather than planning for widespread job replacement, companies may find greater value in ensuring that employees understand when and how to use AI effectively in their daily responsibilities.
As AI capabilities continue to develop, workplace adoption is likely to depend not only on the technology but also on how organisations prepare employees to work alongside it.
What the findings mean for workplace AI strategies
The findings suggest organisations should treat AI as a productivity tool rather than a replacement for workers.
The largest gains are likely to come from using AI to automate routine tasks, improve decision-making and help employees focus on higher-value work.
For companies, this means moving beyond simply giving employees access to AI tools and developing a structured strategy for their use.
Organisations will need to identify areas where AI can deliver measurable improvements, train employees to use the technology effectively and establish clear guidelines for responsible adoption.
Without adequate planning, businesses risk investing in AI without achieving meaningful returns.
The findings also highlight the need for workforce readiness.
As AI becomes part of everyday work, employees will need new skills to use these tools effectively. Companies that invest in AI literacy and upskilling are likely to gain the most from the technology while easing the transition for workers.
Ultimately, the report points towards a future in which AI acts as a productivity partner rather than a replacement for workers. Workplace AI strategies will need to balance technological investment with human expertise.
The broader picture
The growing use of AI in workplaces shows that the technology is moving beyond simple automation and becoming part of everyday business operations.
Companies are increasingly using AI to help employees analyse information, improve workflows, generate ideas and handle repetitive work.
The key takeaway is that AI’s value will come not only from replacing manual processes but also from helping people work faster, make better decisions and focus on more meaningful tasks.
For businesses, the challenge is to ensure that AI adoption is practical and responsible.
Simply introducing AI tools will not guarantee better results. Organisations will need to train employees, establish clear usage policies and understand where AI can have the greatest impact.
As the technology evolves, companies that combine AI capabilities with human skills will be better placed to improve productivity.
 "Google study finds workplace AI acting more as assistant than replacement")
