
The Department of Philosophy Colloquium Series presents:
"When to Swipe Left on AI: Testing LLMs against Aristotle鈥檚 criteria for true friendship"
Speaker: Jason Breen (Department of Philosophy)
Monday, Nov. 24聽 | 12:30 - 2 p.m.
免费福利资源在线看片 Hall聽B716
Abstract: Aristotle writes that 鈥渋t is not possible to have friendship toward inanimate things鈥 (NE VIII.2, 1155b27鈥33). Yet when Replika removed its romantic role-play features in 2023鈥2024, thousands of users reported grief and disorientation, describing the change as 鈥渓osing a partner overnight.鈥 What are we to make of this apparent contradiction? Humans clearly form emotional attachments to chatbots, but can these chatbots truly be considered friends? If so, in what ways can they fulfill that role?
I argue that AI companions such as Replika may simulate some forms of friendship, but they cannot satisfy Aristotle鈥檚 criteria聽for genuine friendship because they lack intrinsic valuation and the capacity for reciprocal concern. Transformer architectures can simulate attention, but reinforcement learning grounds only instrumental optimization, not the mutual recognition required for friendship.The stakes extend beyond personal wellbeing. For Aristotle, friendship is also a civic virtue that sustains democratic life through reciprocity, shared recognition, and co-deliberation. When citizens form attachments to entities that cannot reciprocate, they may habituate to one-sided, low-demand relationships. This weakens the dispositions needed for civic friendship and encourages governments and platforms to substitute cheap AI solutions for more human and expensive forms of civic education and participation.
Contact:
David Balcarras | david.balcarras@uleth.ca | (403) 329-2462