William Stanley Jevons' paradox suggests that increased efficiency in resource use leads to greater overall consumption. This principle, initially applied to coal, now extends to judgments as artificial intelligence, like TypeSafe's Jev, makes individual judgments cheaper. When the cost of a judgment drops significantly, the total volume of judgments made is expected to explode. This trend is exemplified by ATMs, which automated routine transactions but shifted tellers' roles to more complex customer service and sales. Automating the "easy" judgments does not shrink a category but rather re-sorts it, leaving behind the more difficult cases.The judgments that are automated tend to be those with defined options, recoverable errors, downstream checks, and high repetition. Crucially, these are judgments where the cost of error is borne by a system rather than by an individual's reputation or finances. The boundary of automatable judgments is defined by where the consequences of being wrong land. Consequently, when the cheaper, system-borne judgments are automated, the remaining human-assigned judgments are characterized by direct accountability.The paradox highlights that cheapening inference, a computational component of judgment, increases the overall volume of decisions. While the cost per judgment and time-to-verdict fall, the cost of owning the outcome remains constant. Accountability, unlike inference, has no efficiency curve and cannot be optimized. Therefore, as inference becomes nearly free, the difficulty of a decision shifts to the person who must ultimately own its consequences. The most expensive judgments are those where multiple outcomes are defensible and a person must bear responsibility.