TL;DR

OpenAI has published its principles for managing the “capability overhang”—the gap between what AI can do and what people actually use it for. Data reveals power users leverage 7x more AI capability than typical users, suggesting significant untapped potential.

The Capability Overhang

Three years after ChatGPT’s launch, OpenAI has shared its perspective on a critical challenge: most people, businesses and countries are not capturing the full value that AI systems can now deliver. The company calls this the “capability overhang”—a gap that presents both an opportunity and a responsibility.

OpenAI’s usage data reveals a striking disparity: typical ChatGPT power users consume seven times more compute power than average users. These power users don’t simply use AI more frequently—they apply advanced capabilities across a wider range of tasks, producing more economically valuable work as a result.

Three Principles for Managing AI Adoption

OpenAI has outlined three principles guiding their approach:

First to Truth: In periods of rapid change, accurate information creates agency. OpenAI commits to sharing foundational economic insights, including measurements of how AI tools compare with human performance across various tasks.

Access: As AI shapes economic outcomes, access to core tools becomes essential. This principle underpins OpenAI’s decision to maintain a free tier of ChatGPT, now supported by advertising, and to provide APIs for developers building applications.

Self Empowerment: Tools should enable people to shape their future in ways the creators didn’t anticipate. This applies to everyday users optimising family budgets or job searches, and to founders turning frontier capabilities into new companies.

Looking Forward

For UK SMEs, this framework offers practical guidance: the businesses capturing most value from AI are those exploring capabilities comprehensively rather than superficially. OpenAI’s vision of “double-digit GDP growth” and “rapid scientific advancement” may seem ambitious, but the underlying message is clear—the gap between AI potential and actual adoption represents a competitive opportunity for those willing to invest in learning these tools deeply.