CONCEPT
The Underdetermination Problem
The structural condition in which empirical evidence alone cannot settle disputes between well-developed theoretical traditions — because the traditions differ not only in predictions but in what counts as evidence and what counts as disconfirmation.
Underdetermination is the condition that makes disputes between research traditions interminable by empirical means alone. The Duhem-Quine thesis established the general principle: any theoretical claim, taken in isolation, can be reconciled with any empirical observation by adjusting auxiliary hypotheses. Laudan extended the principle to research traditions as a whole. When traditions disagree about what counts as relevant evidence, what predictions their theories generate, and what observations would count as disconfirming, no single experiment — however rigorous — can resolve the dispute. The traditions can absorb the same evidence into opposed interpretations. The resolution of such disputes requires not a crucial experiment but sustained accumulation of problem-solving success across multiple domains over extended periods.
In The You On AI Field Guide
Underdetermination is the structural feature that makes the Berkeley study simultaneously cited by the triumphalist tradition as evidence of productivity gains and by the elegist tradition as evidence of work intensification. The study documented that AI-assisted workers worked