Anthropic Study: AI's Job Automation Potential Far Exceeds Real-World Use

• Anthropic study reveals a significant gap between AI's theoretical job automation capabilities and its actual workplace usage.
• While computer and mathematics roles show 94% theoretical exposure, actual AI usage in these fields is only 33%, with most sectors below 20%.
• "Observed exposure" highlights computer programmers (75%) and data entry (67%) as most impacted by AI automation, not just assistance.
• The study challenges assumptions, finding vulnerable workers are often older, highly educated, female, and higher paid, not primarily blue-collar.
• Despite rapid AI capability expansion (49% of US jobs now have >25% tasks exposed), upskilling is key as AI reshapes roles rather than eliminating jobs.