Epoch AI

Epoch AI was founded to help people understand what is happening in AI from a neutral perspective and grounded in the best possible evidence.

Our goal is to improve society’s understanding of the drivers, progress, and impact of artificial intelligence. We focus on building a shared scientific foundation for thinking about AI, neutral to any specific agenda, so decisions about this technology are informed by the best available data.

We examine the driving forces behind AI, track the progress of AI capabilities, and measure and forecast its impact. We ground all our research in data and distill it into accessible tools and explainers.

Drivers

We investigate the drivers and bottlenecks of AI progress and scaling. We were among the first to systematically monitor trends in AI training compute, and to identify crucial shifts such as the role of inference compute scaling. Our research includes authoritative analyses of whether AI scaling can continue through 2030 and the feasibility of decentralized training, providing concrete insights into what constrains AI progress.

We develop and host leading databases tracking the inputs to AI development, including ML hardware, AI model training compute estimates, investment, GPU clusters, and frontier data centers. This data provides the foundation for understanding AI’s trajectory and resource requirements.

Progress

We measure and track the capabilities of AI systems over time. We regularly run independent evaluations of new models across domains, and go beyond surface metrics to understand concrete implications for model capabilities. We also review prominent benchmarks, and create new ones; for example, FrontierMath, the state-of-the-art for testing frontier math capabilities.

Impacts

We gather evidence and develop scientifically grounded models to forecast the impact of AI on the economy and the world at large.

We have conducted highly cited research on the economics of AI, from analyzing the consequences of automating remote work, to quantitatively modeling the impact of AI development on the economy. We also conduct polls and interviews on how AI is used and perceived by academics, AI researchers, and the general population.