Google's Analysis of 15 Million Real AI Interactions Finds Most Work Unchanged
The short answer
A Google study of 15 million real-world AI interactions found that most tasks at most jobs remain unaffected by AI tools — one of the largest empirical looks at how workers actually use AI day-to-day. If your team hasn't seen dramatic productivity gains from AI yet, the data suggests that's the norm rather than an exception.
- AI adoption
- productivity
- workforce
- ROI
What happened
Analysis of 15 million real AI interactions conducted by Google finds that most tasks at most jobs are not being automated, according to Ars Technica. The data provides one of the larger empirical looks at how workers use AI in practice, and the picture it reveals is more restrained than the surrounding discussion would suggest. The headline finding: most job tasks, across most roles, are simply unaffected.
The results do not indicate AI creates no value, but they do point to uneven adoption where impact concentrates in specific task types rather than spreading evenly across entire job functions.
What it means for your business
If your team hasn't seen dramatic productivity gains from AI yet, the data suggests that's the norm rather than an exception. The evidence points toward focusing on specific, high-volume task categories — drafting, summarizing, research — rather than expecting blanket efficiency improvements across every role. For small businesses evaluating whether an AI subscription is earning its monthly cost, that targeted framing is useful: track whether it actually saves time on the tasks you assigned it to, not whether it transformed how the whole team works. A company-wide AI mandate is harder to justify than a narrow deployment with a measurable outcome.
Source: Ars Technica