본 연구는 의인화 이론을 토대로 인공지능이 미술작품을 창작하였다는 정보가 감상자의 작품 평가와 창작자 마음인식에 미치는 영향을 분석하고자 하였다. 특히 인공지능에 대한 태도의 조절효과를 고려하여, 응답자가 인공지능에 대해 긍정, 중립, 혹은 부정적인 경우 인공지능 창작자 효과가 어떻게 달라지는지 비교하였다. 온라인 실험에서 참여자들은(N = 349) 인공지능이나 인간 예술가가 창작하였다고 소개된 그림을 감상한 후 창작주체의 마음과 그림의 작품성을 평가하였다. 실험 결과 인공지능에 부정적인 응답자는 창작주체에 따른 작품 평가에 차이가 없었던 반면, 긍정적인 응답자들은 인공지능 작품의 예술성을 더 높게 평가하였다. 또한 인공지능에 부정적인 응답자들은 인간 창작자에 비해 인공지능에 대한 마음인식이 낮은 반면, 긍정적인 응답자들의 그 차이가 작거나 유의미하지 않았다. 인공지능에 대해 부정적인 응답자에게서 마음인식은 인공지능 창작자 효과를 부적으로 매개하였으나, 긍정적인 응답자에게 마음인식의 매개효과는 유의미하지 않았다. 이는 일반적인 사람들은 인공지능으로부터 창작의도나 감정과 같은 마음상태를 인식하지 못해 인공지능의 작품을 낮게 평가하지만, 인공지능에 긍정적인 사람들에게는 마음인식 이외의 경로로 인공지능 작품의 평가가 높을 수 있음을 시사한다.
As Information and Communication Technologies (ICTs) have advanced, artificial intelligence (AI) has been used in broader domains, including artwork creations. It is believed that creating artwork belongs only to humans, as they are only able to appreciate the true aesthetic meanings of artworks. Yet, previous studies suggest that people often consider artworks created by AI as valuable as those created by human artists. To this end, the current study aimed to investigate how the information that AI is a creator of an artwork influences individuals evaluating the artwork. For this purpose, we examined individuals’ responses across different attitudes toward AI. Literature on anthropomorphism suggests that people employ human knowledge, such as how people look or feel in a certain situation, to infer the invisible natures of an unknown target. This argument suggests that people would use knowledge of human artists to interpret an AI artist, when they are informed that artwork is created by the AI creator. This interpretation, we expect, would be influenced by individuals’ attitudes toward AI, in general. Thus, we hypothesized that people would evaluate AI-created artworks as valuable as human-created artworks, if they are positive toward AI. We also developed hypotheses on how people perceive mental states from an AI creator differently from those from a human creator, such that people perceive experience (i.e., a mental capability of feeling emotion) and cognition (i.e., a mental capability of planning and forming intention) from an AI creator as much as a human creator when they are positive toward AI, but to a greater extent from a human creator when they are negative toward AI. To test the hypotheses, we conducted a one-way online experiment (N = 349), in which participants were exposed to paintings reportedly created either by an AI or by a human. After they watched two paintings, they rated each artwork and the creator’s mind. The differences in their ratings between the AI creator and human creator were compared for respondents with negative, neutral, and positive attitudes toward AI. As predicted, the differences were contingent upon the participants’ attitudes. Results revealed no differences in art evaluation for negative- or neutral-attitude participants, whereas AI-created paintings were rated higher than human-created paintings by positive-attitude participants. Also, negative- or neutral-attitude participants attributed experience and cognition to a lesser degree to the AI artist than to the human artist, whereas the difference was not significant for positive-attitude participants. Also, our results evidenced that mind attribution mediated the effect of the AI artist on artwork evaluation in a negative direction for negative- or neutral-attitude participants, but it was not significant for positive-attitude participants. These results indicate that people who were negative toward AI may devalue artworks created by AI, possibly because they are less likely to ascribe mental states to an AI artist and aware that an AI artist lacks of mental abilities to produce valuable artwork. People who are positive toward AI, however, may fully appreciate the aesthetic values of AI artworks to a greater extent than those from human-created artwork.