Many questions regarding Artificial Intelligence (AI) systems reside in the domain of the nature of their processes and how much these processes are similar to human cognition. These questions are central to our understanding of AI’s growing presence in our everyday activities. Of special interest is the recent advancement with Large Language Models (LLMs). Due to their vast use of text data, they seemingly produce linguistic behavior that is similar to, and often indistinguishable from, our own. In this paper, we draw on the enactive theory of mind and language to re-situate the traditional problems of meaning as the mind-in-life problem for AI. The enactive theory posits that meaning is given by sense-making, an ability of living systems due to their precariousness, autonomy, self-maintenance, and self-production. We suggest that AI systems cannot be sense-making systems because they lack the fundamental biological structure that enables sense-making. Therefore, although the AI’s use of symbols can be said to be operationally efficient, it is not truly meaningful.
Rolla et al. (Thu,) studied this question.