This paper reports two studies examining participants' identification accuracy in discriminating real faces from realistic “artificial” faces created through the Artificial Intelligence (AI) system StyleGAN. Across the two studies, two different sets of participants (N = 400) attempted to distinguish 24 real from 24 AI-generated images. Both sets of participants exhibited poor discrimination accuracy and a bias to report all images as real (Study 2). We examined other possible influencing factors were examined, such as smile intensity (Study 1) and age-congruence between participants and faces (Study 2). Implications for future research, and for understanding the potential societal impacts of AI-generated online content are discussed.