AI Briefing
KO

Stanford AI Index 2026 - AI Is Sprinting, and We're Struggling to Keep Up

·2026.04.16 18:54

Key point

An annual report condensing the latest indicators of AI adoption, performance, investment, jobs, and trust.

Details

Stanford HAI has released the 423-page AI Index 2026. The core message is that AI is advancing very rapidly in terms of performance, adoption, and investment, while social adaptation is struggling to keep pace.

Performance and Adoption

  • The global adoption rate of generative AI reached 53%, spreading faster than PCs and the internet.
  • Frontier models have surpassed humans on PhD-level science problems and Math Olympiad-level tasks.
  • The real-world task success rate of AI agents rose sharply from 20% → 77.3% in just one year.
  • However, reading analog clocks remains stuck at just 50.1%, a large gap from the human average of 90%.

US-China Competition

  • Anthropic ranks 1st, with the gap to Chinese models at just 2.7 percentage points, effectively close.
  • US private AI investment stands at $285.9 billion, versus China's $12.4 billion, a roughly 23x difference.
  • Meanwhile, the number of AI researchers migrating to the US has dropped 89% compared to 2017.

Jobs and Productivity

  • Employment of software developers aged 22-25 has fallen 20% compared to 2024.
  • Customer support productivity rose 14%, and software development productivity rose 26%.
  • One-third of companies expect AI-driven workforce reductions within the next year.

Environmental Impact

  • Carbon emissions from training Grok 4 reached 72,000 tons, 14 times that of GPT-4.
  • GPT-4o's annual water usage was cited as equivalent to the drinking water needs of 12 million people.

Trust and Perception

  • Trust in US government AI regulation stood at 31%, the lowest among surveyed countries.
  • Gen Z's "anger toward AI" rose to 31%, up from 22% last year, while "excitement" fell to 22%, down from 36% last year.
  • There was a 50-percentage-point perception gap between AI experts (73%) and the general public (23%) regarding AI's impact on jobs.
  • The number of AI-related incidents rose from 233 in 2024 to 362 in 2025.

The original report is linked from Stanford HAI's AI Index 2026 page.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.