Creating Artificial Intelligence

Abstract

Through an inductive, qualitative study of individuals developing new artificial intelligence (AI) technologies, this dissertation builds theory on how creative workers manage the emotions that arise from forecasting the outcomes of implementing their creations. I find that, in a context that illuminates the danger of implementing certain types of creative ideas, creative workers forecast both positive and negative outcomes arising from implementing their work, which elicits ambivalence. My work indicates that how creative workers respond to this ambivalence affects whether they impose constraints on their work as it unfolds. First, some individuals may proceed without constraints because they have resolved their ambivalence by amplifying their positive thoughts and feelings toward their work. Informants who exhibited this pattern created psychological distance (Lewin, 1951; Trope & Liberman, 2003) from the potential negative effects of their work by anchoring on the present moment and/or emphasizing potential positive outcomes. However, the majority of informants exhibited a novel “redistribution” response to ambivalence, whereby they committed to their work (Brickman et al., 1987; Pratt & Rosa, 2003; Pratt & Pradies, 2011) and shifted from a strengthening of negative thoughts and feelings toward a strengthening of positive thoughts and feelings through the use of self-imposed constraints. My work suggests that, although self-imposed constraints do not eliminate negative thoughts and feelings altogether, applying these self-determined boundaries enables individuals to reduce ambivalence and engage (Harter, Schmidt, & Hayes, 2002) more fully in their work. In addition to inducing a process model that encompasses these dynamics, I present the categories and types of self-imposed constraints that I have induced. These self-imposed constraints are not mutually exclusive, and each serves one of three broader purposes: developing a sense that one’s creation will have a positive moral valence, that one will be able to control his or her creation, or that one may trust in the quality of his or her creation. This dissertation extends theory on the role of prospective thought processes in creative work and shows how constraints, though often seen as impediments to creativity, can be used proactively by creative workers to manage the darker emotions and thought processes that have largely been overlooked in prior research. This work also contributes a novel response to ambivalence, redistribution, which entails approaching potentially harmful creative work in a heedful manner.