Teleological Explanations

Teleological explanations explain a phenomenon by reference to its purpose, function, or end-state — rather than by reference to prior causes. To say that the heart beats in order to circulate blood is a teleological explanation; to say the heart beats because of electrical signals from the sinoatrial node is a mechanistic explanation. Both can be true simultaneously. The question is whether teleological explanations are merely heuristic shorthand for mechanistic ones, or whether they pick out something real that mechanistic accounts cannot capture.

In evolutionary biology, teleological language is ubiquitous: organisms 'try' to survive, adaptations 'serve' functions, genes 'seek' replication. The standard naturalistic teleology interprets these claims as claims about selected-for causal histories: a trait has the function of X if it was selected because it produced X. This strips teleology of literal intentionality while preserving the functional vocabulary. Whether it preserves everything that matters — whether the history-dependence of biological function captures the normativity that function-talk seems to carry — is contested in the philosophy of biology.

The import for AI is acute: if AI systems optimize for objectives, are they exhibiting teleological behavior in any meaningful sense, or merely simulating the grammatical surface of purposive language? A system that maximizes reward is not trying to maximize reward in the way an organism intends an outcome — or is it? The category error to avoid is assuming that because the functional description fits, the intentional description must too. Or, equally, assuming that because the system lacks biological history, the functional description cannot be literal.