CONCEPT
Polanyi's Paradox
The observation that humans know far more than they can articulate—and that this <em>tacit knowledge</em> has historically resisted automation, predicting the difficulty of computerizing adaptive judgment.
Polanyi's Paradox, named by MIT economist David Autor in 2014, formalizes Polanyi's insight that "we can know more than we can tell" into an economic principle governing automation. The paradox explains why certain tasks—diagnostic reasoning, skilled craftsmanship, contextual judgment—proved stubbornly resistant to computerization even as routine cognitive work was automated. These resistant tasks require tacit knowledge: embodied sensitivities built through practice that operate below conscious awareness and resist specification in algorithms. Autor's framework predicted that AI would struggle most with precisely these tacit-knowledge-intensive domains. The 2020s arrival of large language models appeared to overcome the paradox through statistical pattern-matching at scale, but critics argue this represents "Polanyi's Revenge"—systems that capture patterns without understanding produce new categories of failure invisible to their operators.
In The You On AI Field Guide
Autor developed the paradox while analyzing why computerization had not produced the mass unemployment economists predicted in the 1960s. He found that computers automated tasks that could be specified in logical rules while creating demand for tasks requiring flexibility, judgment, and contextual