Why AIs Cannot Play Games

Abstract

This paper explores the human experience of game-playing and its implications for artificial intelligence. The author uses phenomenology to examine game-playing from a human-centered perspective and applies it to language games played by artificial intelligences and humans. The paper argues that AI cannot truly play games because it lacks the intentionality, embodied experience, and social interaction that are fundamental to human game-playing. Furthermore, current AI lacks the ability to converse, which is argued to be equivalent to Wittgenstein’s view of engaging in “language games.” The author presents a formal ontology of game playing, agents, and their intentional states, and contrasts human agents playing games with artificial agents “playing games.” The paper concludes that while artificial intelligence can simulate some aspects of human language games, it lacks the essential capacities for game playing, such as cognition, creativity, will, intentionality, understanding, and personhood

Author's Profile

David Koepsell
Texas A&M University

Analytics

Added to PP
2024-06-15

Downloads
68 (#74,160)

6 months
68 (#93,074)

Historical graph of downloads since first upload
This graph includes both downloads from PhilArchive and clicks on external links on PhilPapers.
How can I increase my downloads?