Dialog-IA: A Debate on Artificial Intelligence (AI) from Dialogic Discourse Theory
TL;DR
The ethical limits of AI, particularly generative AI, such as ChatGPT, are discussed, questioning its impact on human processes of interaction and meaning-making, and the extent to which AI can engage in authentic dialogue is assessed.
Abstract
ABSTRACT In this article, we aim to discuss, based on Bakhtinian dialogic theory, the ethical limits of AI, particularly generative AI, such as ChatGPT, questioning its impact on human processes of interaction and meaning-making. To this end, we present a historical overview of the origins of AI as a process embedded in transformations in social relations. Next, grounded in Bakhtinian theoretical principles of dialogue, responsiveness, and ideology, we assess the extent to which AI can engage in authentic dialogue. With our discussion, we highlight, on the one hand, that in Bakhtinian terms, the discursive horizon of AI is limited to the corpus with which it was trained, blocking the authentic responsive dialogic nature of this interaction. On the other hand, we emphasize that this limitation - determined by AI creators - reproduces cultural and ideological patterns, potentially silencing various social voices. Finally, we stress the need to consider that AIs respond to the context of neoliberal transformations in production and consumption relations, characterized by the overlap between media, market, and the commodification of information. In light of this, it is ultimately the human being who must ethically assume responsibility for decision-making in the use and creation of technologies.