Stanford University has released the third edition of its AI100 report, titled AI Goes Mainstream: The Scaling Era and the Rise of Generative AI. The study, which covers the period from 2021 to 2026, analyzes the rapid integration of generative AI into global public life, science, and labor markets.
The report identifies that recent technological breakthroughs were primarily driven by increased computational power and larger training datasets. However, the authors warn that future progress faces significant obstacles, specifically the rising costs of compute and the limited availability of high-quality training data.
A key finding highlights that the development of frontier AI is currently concentrated among a small number of organizations based in the United States and China. The committee suggests that this concentration poses governance challenges and advocates for increased public involvement in deployment decisions.
The AI100 project, launched in 2014, aims to produce assessments every five years over a century-long span. The panel, which includes experts from computer science and philosophy, utilized AI tools for tasks such as transcription and drafting, while maintaining human oversight for core analysis.
