Intentionality and Multiple Agents

In the AI Story

The intentionalist framework had a specific methodological attraction for twentieth-century editorial theory: it provided a principle for deciding between textual variants. When two versions of a sentence exist, the editor privileges the one that represents the author's final intention. This required, for its operation, the assumption that authorial intention was recoverable and that it was methodologically prior to other kinds of intention.

McGann's challenge was that this assumption was ideological rather than factual. Authorial intention, as a recoverable object, does not exist in the simple form the framework requires. Every act of authorial writing is embedded in a social network that shapes what the author can intend. The poem Byron published was shaped by his anticipation of Murray's responses, his awareness of reviewer sensitivities, his calibration to audience expectations. The 'author's intention' was always already responsive to the intentions of other agents. Separating it out is a theoretical operation, not a factual recovery.

Applied to AI collaboration, the framework handles the intentionality question with specific precision. Claude does not have intentions in the sense that a human editor does. But the AI's outputs shape what the author's text becomes, and the shaping is not fully attributable to the author's prior intention. The structural framework Claude suggests, the connection Claude draws, the passage Claude generates that the author accepts — these contributions alter what the text means, and the meaning that results belongs to the collaboration rather than to either participant's prior intention.

This is not a dissolution of authorship into meaninglessness. The author's intention is real and genuine; it is simply not the sole determinant of textual meaning. The author's intention interacts with the collaborator's capabilities, and the outcome is something neither would have produced independently. McGann's framework accommodates this without difficulty because it was never committed to the simple attribution model the intentionalist tradition required.

Origin

The argument was developed in A Critique of Modern Textual Criticism (1983) and refined across The Textual Condition (1991). It drew on earlier critiques of intentionalism in literary theory, most notably W.K. Wimsatt and Monroe Beardsley's 'The Intentional Fallacy' (1946), but extended those critiques into the specific domain of textual editing.

Key Ideas

Intention as input, not origin. Authorial intention is one input to textual production, not the sole origin of textual meaning.

Ideological premise. The intentionalist framework was a theoretical construction rather than a factual recovery; it required assumptions that evidence does not support.

Social embedding of authorial intention. Every act of authorial writing is responsive to the intentions of other agents in the social network of production.

Emergent meanings. Interactions between multiple intentions produce meanings that no individual agent intended, and these emergent meanings are part of what the text means.

AI's compatibility. The framework accommodates AI collaboration without requiring new theoretical apparatus; the AI is another participant in a network that has always been multi-agent.

Debates & Critiques

The intentionality debate has been active in literary theory and editorial practice for most of the twentieth century. McGann's position has largely prevailed in textual scholarship, though some editors continue to defend intentionalist practice for specific projects. The debate continues in philosophy of mind, where the question of whether AI systems have intentions in any meaningful sense remains contested.