Training Response-Able Machines
Jamie Jelinek · 2024
In November 2023, in a blog post celebrating the ten-year anniversary of their Fundamental Artificial Intelligence Research (FAIR) team, Meta explains that the future of the team will focus on building “socially responsible AI.” Rhetorical scholarship has demonstrated how automated technologies exert agency and power within digital networks, but what does it mean for a machine to act responsibly? This chapter analyzes Meta’s descriptions of their “Responsible AI” initiative and demonstrates how Meta positions their AI as an autonomous rhetorical agent with its own machinic ethos. The use of training to shape ethos is a method that mirrors the rhetorical techne of Isocrates. By instilling the habits of kairos and to prepon into their AI models, this chapter argues that the “responsible AI” described by Meta is trained through machine learning to display a machinic ethos of response-ability. Focusing on response-ability adds an important nuance to the larger project of Responsible AI because it asks who is responsible for the way a machine is taught to respond.