The AI Index 2025, one of the most complete and respected reports on the evolution of artificial intelligence in the world, prepared by Stanford University.
The publication shows that we are no longer talking about the future — AI is shaping the present, with a direct impact on science, the economy, education, and the way businesses operate.
Below, I highlight some of the main points:
Global Panorama
Private investment in AI reached $252 billion in 2024, with the US leading, followed (at a distance) by China and the UK. Open models are increasingly closer to closed ones—the gap has narrowed from 8% to just 1,7% in benchmarks. Corporate adoption has exploded: 78% of companies reported using AI in 2024 (down from 55% in 2023). The cost of model inference has fallen more than 280-fold in just 18 months.
Technical Advances
Models have surpassed challenging benchmarks (GPQA, SWE-bench) with leaps of up to 67 percentage points. Small models (like Microsoft's Phi-3-mini) are already surpassing results previously restricted to giants with billions of parameters. The new frontier includes realistic video generation, human-rivaling agents, and new reasoning approaches.
Accountability and Governance
AI incidents grew 56,4% in 2024. Public trust in AI companies fell from 50% to 47%. Fifty-nine new federal regulations related to AI were implemented in the US (more than double the number in 59). Countries like France, China, Saudi Arabia, and India are creating megaprojects for AI infrastructure and governance.
Science and Medicine
Models have outperformed doctors in complex clinical tasks and diagnoses. The number of FDA-approved AI medical devices has jumped from 6 (in 2015) to 223 (in 2023). The Nobel Prizes in Chemistry and Physics have awarded work directly linked to AI.
Education and Public Opinion
Two-thirds of countries already offer (or plan to offer) computer science education in basic education. Global optimism about AI rose from 52% to 55%, with countries like China (83%) and Indonesia (80%) standing out. In the US, only 13% of people trust self-driving cars—distrust still hinders many applications.
My reading? AI has established itself as critical infrastructure—economic, scientific, and strategic. The numbers don't lie: anyone not creating real value with AI by 2025 risks being left behind.
Are you already exploring the report's data? Here's the link for anyone who wants to delve deeper: https://aiindex.stanford.edu/report/
