Deflating Deflationism: A Critical Perspective on Debunking Arguments Against AI Mentality

Abstract: Many people feel compelled to interpret, describe, and respond to Large Language Models (LLMs) as if they possess inner mental lives similar to our own. Responses to this phenomenon have varied. \textit{Inflationist} views endorse the truth of such ascriptions, granting that at least some attributions of mentality to LLMs are warranted. \textit{Deflationists} instead are more sceptical of these attributions, often cautioning against the risk that anthropomorphic projection may lead to misplaced trust or potentially even confusion about the moral status of LLMs. We advance this debate by assessing two common deflationary arguments against LLM mentality. What we term the \textit{robustness strategy} aims to undercut one justification for believing that LLMs are minded entities by showing that putatively cognitive and humanlike behaviours are not robust, failing to generalise appropriately. What we term the \textit{etiological strategy} undercuts attributions of mentality by challenging naive causal explanations of LLM behaviours, offering alternative causal accounts that weaken the case for mental state attributions. While both strategies offer powerful challenges to full-blown inflationism, we find that neither strategy provides a knock-down case against ascriptions of mentality to LLMs \textit{simpliciter}. With this in mind, we explore two modest forms of inflationism about LLM mentality that permit ascriptions of mentality to LLMs under certain conditions.\textit{ Practical modest inflationism }holds that we can, and perhaps should, mentalise LLMs where it is practical to do so, provided the benefits are weighed against relevant risks including the risk of potentially problematic forms of emotional dependency.\textit{ Metaphysical modest inflationism} holds that we can permissibly attribute those mental states and capacities which can be understood in metaphysically undemanding terms (such as knowledge and belief) while exercising greater caution when attributing more metaphysically demanding mental phenomena such as consciousness.