AI and Strategic Judgment

The practical question is not whether AI is generally “good” or “bad” at strategy. It is which cognitive activities fit the architecture of a particular system, under which conditions, and with which forms of human oversight.

Large language models provide extraordinary fluency, broad association, and rapid recombination. Those capabilities can create real leverage in search, articulation, comparison, translation, and the development of candidate explanations. They do not automatically supply causal understanding, reliable relevance selection, genuinely novel theoretical structure, or responsibility for action.

This program joins conceptual analysis to empirical evaluation. It examines strategic tasks, forecasting, point-in-time testing, model portfolios, agent and harness interaction, and the conditions under which human–AI complementarity improves judgment rather than producing cognitive surrender.