The qualification problem is the difficulty that, when you try to state the conditions under which an action will succeed, you can never finish: there is always one more qualification you have not listed. John McCarthy named and described it at the start of his 1980 paper on circumscription, cited here. His example is using a boat to cross a river. The boat will get you across only if it is not leaking, the oars and rowlocks are present and unbroken and fit each other, there is no hole, no one has chained it up, and so on without obvious end. Anyone can always think of a further requirement not yet stated.
The problem matters because formalizing common-sense knowledge in logic seemed to demand spelling out every such precondition, which is both impractical and impossible to complete. McCarthy distinguishes it from the closely related frame problem, which concerns stating everything an action leaves unchanged; the qualification problem concerns stating everything that must hold for an action to work at all. Both arise from the same root, the impossibility of writing down exhaustively explicit knowledge about an open-ended world.
McCarthy’s proposed remedy was nonmonotonic reasoning, specifically circumscription: instead of listing every qualification, the reasoner assumes the action works unless something known prevents it, treating obstacles as exceptions to be minimized. STRIPS-style planners sidestepped a related version of the difficulty by restricting how actions could be described. The qualification problem nonetheless remains a touchstone in discussions of common-sense reasoning and the brittleness of rule-based systems.
Why a business reader should care: any attempt to automate decisions with hard-coded rules eventually collides with the qualification problem, because the real world keeps supplying edge cases the rules never anticipated, which is precisely why robust systems need ways to act on sensible defaults rather than complete specifications.