A large analysis of AI use suggests that most workers are not handing over most of their jobs to automation.
Ars Technica reports on Google data covering 15 million real AI interactions. The analysis found that most tasks in most jobs are unaffected, even as AI tools become more common in daily work.
The finding cuts against the simplest versions of both AI hype and AI panic. Models can help with writing, coding, summarizing, and analysis, but many jobs include physical tasks, interpersonal judgment, domain-specific responsibility, or messy workflows that are not easily automated end to end.
The practical implication is that adoption may look uneven. Some tasks will change quickly, while entire roles change more slowly. Employers and workers still need to prepare for disruption, but the data suggests the near-term story is task reshaping rather than instant replacement across the labor market.