2024-03-24-NBER-AI_ADOPTION_IN_AMERICA-_WHO,_WHAT,_AND_WHERE_66页_1mb

AI Adoption in America: Who, What, and Where

1. Technology Adoption Rates:
   Only ~6% of U.S. firms adopted at least one AI-related technology in 2017. However, AI use was highly concentrated among very large firms (over 5,000 employees), yielding an employment-weighted adoption rate of 18%. Specific AI technologies like machine learning (2.9%) and machine vision (0.8%) had lower individual adoption rates compared to the combined measure.

2. Industry Patterns:

   • Leading Sectors: Manufacturing (12%) and information (12%) were the top industries for AI adoption.
   • Emerging Tech Hubs: Medical, diagnostic laboratories, software publishing, and computer systems design showed high AI prevalence (~23%–16%).
   • Lagging Sectors: Retail trade and construction (~4%) lagged behind, while FIRE (Finance, Insurance, Real Estate) sectors had low adoption despite high digitization.

3. Adoption Correlates:

   • Venture Capital: VC-funded startups were 2.9% more likely to use AI.
   • Innovation: Firms with recent process or product innovation (39% and 66% respectively for AI users) and formal IP reliance (5.2% ownership/pending patents) showed stronger AI adoption.
   • Geographic Hubs: AI use was geographically concentrated in major metropolitan areas, particularly in the West and South (e.g., Nashville, Las Vegas, Tampa). Employment-weighted adoption identified emerging hubs in the Midwest/Mid-Atlantic, including Louisville, Columbus, and Austin.

4. Entrepreneurship and Ownership:

   • Founder Characteristics: AI-using companies had startup owners with higher education, prior business experience, and a mix of growth-oriented (61.8%) and prosocial (30.3%) motivations.
   • Age Factor: Younger owners (<35) had slightly higher AI adoption.
   • Lifestyle Motivations: Lifestyle-focused startups (e.g., flexible hours) showed less AI adoption.

5. Future Trajectory:
   Low average adoption may indicate barriers to AI implementation, but early adopters are scaling up. Persistent disparities could widen the "AI divide," while clustered technology use suggests reinforcing growth in key sectors.