Univocity Without Ontological Identity: An Ockhamist Framework for Human and Artificial Intelligence

(I have tried to think through a very contemporary philosophical issue from the standpoint of a medieval Ockhamist.)

The question at issue is how we are to understand and justify the use of the same terms—such as “intelligence” and “thought”—when applied to both human beings and artificial intelligence. More precisely, the problem concerns whether such shared linguistic usage commits us to an underlying sameness in what these terms denote, or whether it merely reflects a conceptual practice that does not entail ontological identity. The difficulty lies in distinguishing between the legitimacy of using the same concept across different domains and the unwarranted inference that what is thus described must be the same in kind.

Discussions of univocity and parsimony in William of Ockham do not directly resolve this question. Rather, they may be understood as one among several useful conceptual frameworks for approaching it. In this sense, Ockham’s position does not deliver a final answer, but provides criteria for determining how the question should be formulated and what may legitimately be claimed within it.

The central point in this framework is the meaning of univocity. In Ockham, univocity does not imply the identity of being itself. It concerns, rather, the use of concepts and language. When a predicate is applied to different objects under one and the same concept, it is said to be used univocally. What matters here is the application of an identical concept, not the assumption of an identical ontological status. Univocity is therefore grounded in the identity of the concept formed in the soul.

From this perspective, predicates such as “intelligence” or “thought” may be used univocally of both human beings and artificial intelligence. We identify functional similarities such as problem-solving, prediction, and information processing, and we integrate these similarities under a single concept to which the same term is applied. Such usage is justified insofar as it secures conceptual economy and enables communication.

Yet this univocal usage does not entail ontological identity. Human thought is a complex activity bound up with embodiment, experience, judgment, and explanatory understanding, whereas the operation of artificial intelligence consists in computation, algorithms, and data structures. The two differ fundamentally in their manner of operation and in the way the relevant capacities are realized. It is therefore unjustified to infer from the use of the same predicate that both cases share an identical ontological structure.

At this point, Ockham’s principle of parsimony introduces an important restriction. Similarity between human beings and artificial intelligence may provide the basis for concept formation, but there is no need to posit that similarity as an independently existing real nature. To argue that human beings and artificial intelligence share the same real kind of intelligence simply because the common concept “intelligence” is applicable to both would amount to an unnecessary ontological commitment. Such a commitment ought to be rejected.

Accordingly, similarity should not be understood as an independently existing universal outside the soul, but as a relation grasped in the cognitive process directed toward individual things. When human beings and artificial intelligence produce similar results, we form a common concept on the basis of that similarity. However, there is no need to treat the similarity itself as an independent reality in the external world. The concept exists in the soul, and its application is grounded in cognition of external individuals.

This structure requires that univocity and difference be maintained together. Univocity is secured insofar as the same concept is applied, yet the way in which that concept is realized differs according to the conditions proper to each case. If this point is overlooked, identical linguistic usage may be mistaken for ontological identity, and such a confusion may result in a category mistake.

In the end, Ockham’s position calls for two attitudes simultaneously in any attempt to understand intelligence and thought in human beings and artificial intelligence. The first is to acknowledge conceptual univocity so that the use of the same language remains possible. The second is to preserve the difference in the conditions under which that concept is realized, thereby eliminating unnecessary ontological commitments. These two attitudes do not yield a direct solution, but they provide a rigorous framework for distinguishing what may be said from what ought not to be said in this debate.