Even with superhuman AI, Dario Amodei argues the economic revolution won't be instant. The real-world bottleneck is "economic diffusion": the messy, human process of enterprise adoption, including legal reviews, security compliance, and change management, which creates a fast but not infinite adoption curve.
Related Insights
AI models will quickly automate the majority of expert work, but they will struggle with the final, most complex 25%. For a long time, human expertise will be essential for this 'last mile,' making it the ultimate bottleneck and source of economic value.
The argument that AI adoption is slow due to normal tech diffusion is flawed. If AI models possessed true human-equivalent capabilities, they would be adopted faster than human employees because they could onboard instantly and eliminate hiring risks. The current lack of widespread economic value is direct evidence that today's AI models are not yet capable enough for broad deployment.
Despite the power of new AI agents, the primary barrier to adoption is human resistance to changing established workflows. People are comfortable with existing processes, even inefficient ones, making it incredibly difficult for even technologically superior systems to gain traction.
The AI buildout won't be stopped by technological limits or lack of demand. The true barrier will be economics: when the marginal capital provider determines that the diminishing returns from massive investments no longer justify the cost.
The argument is that "economic diffusion lag" is an excuse for AI's current limitations. If AI models were truly as capable as human employees, they would integrate into companies instantly—far faster than human hiring. The slow rollout proves they still lack core, necessary skills for broad economic value.
For enterprise AI, the ultimate growth constraint isn't sales but deployment. A star CEO can sell multi-million dollar contracts, but the "physics of change management" inside large corporations—integrations, training, process redesign—creates a natural rate limit on how quickly revenue can be realized, making 10x year-over-year growth at scale nearly impossible.
Economist Tyler Cowen argues AI's productivity boost will be limited because half the US economy—government, nonprofits, higher education, parts of healthcare—is structurally inefficient and slow to adopt new tech. Gains in dynamic sectors are diluted by the sheer weight of these perpetually sluggish parts of the economy.
The slow adoption of AI isn't due to a natural 'diffusion lag' but is evidence that models still lack core competencies for broad economic value. If AI were as capable as skilled humans, it would integrate into businesses almost instantly.
AI's "capability overhang" is massive. Models are already powerful enough for huge productivity gains, but enterprises will take 3-5 years to adopt them widely. The bottleneck is the immense difficulty of integrating AI into complex workflows that span dozens of legacy systems.
While AI moves fast in the world of bits, its progress will be constrained in the world of atoms (healthcare, construction, etc.). These sectors have seen little technological change in 50 years and are protected by red tape, unions, and cartels that resist disruption, preventing an overnight transformation.
