Human vs. machine as message source in advertising: examining the persuasiveness of brand influencer type and the mediating role of source credibility for advertising effectiveness in social media advertising

Abstract

Message source effects on persuasion of target audiences have been examined for decades by scholars in advertising, consumer behavior, communication, and psychology, among others. Myriads of studies are available on the subject, but in nearly every instance "source" is defined as a human and rarely is source defined as non-human, or machine. The rise of artificial intelligence (AI) as a message source urges scientific inquiry of the validity of those established theories in a new technology context. The focus of this dissertation is on that of the machine as source. By "machine" this study refers broadly to AI agents, defined to mean digitally created artificial beings that can think and perform tasks like a human. The specific AI agent examined here is that of SMIs, defined as AI agents who are associated with fame and perform human tasks using software and algorithms. The context of the study is social media, defined as "digital networked tools or technologies that enable communication, collaboration, and creative expression across social networks" (McMillan and Childers, 2017, p. 52). Influencer marketing is a crucial component of social media marketing, which is projected to become a $10 billion market by 2023 (Tan, 2019). The primary contribution of this study is, therefore, to understand SMIs' effectiveness in social media advertising. Considering that 95 percent of consumer interaction is projected to be powered by AI by the year of 2025 (Finance Digest, 2020), research to understand the impact of this transformation of message source (from human to machine) is urgently needed but rarely conducted to date. The most apparent machine sources, SMIs, are already being put to use in practice without fully understanding their effectiveness and risk, to replace human influencers. Human influencers, here, specifically refer to social media influencers (SMIs) "who have built a sizable social network of people following them and are seen as self-made microcelebrities" (Shan et al., 2020, p. 2). Indeed, this very notion is reflected in the warnings of the infamous physicist Stephen Hawking who predicted that, someday, machines may even replace humans. As much as this may sound like a futuristic movie, machines are beginning to replace humans in fields as far and wide as medicine, engineering, and transportation. For example, AI is cleaning floors at airports, taking people's temperatures, and even making salad in hospital dining-halls in response to the coronavirus pandemic (Semuels, 2020). In advertising, AI is taking over the work humans have traditionally done, from content matching to advertising creation (Rodgers, 2021). An SMI, Lil Miquela, even takes the spot that is usually reserved for a human and is named as one of Time's 25 most influential people on the internet in 2018 (Time, 2018). The proposed research is supported by survey results from consumers worldwide that paint a mixed picture --some consumers embrace AI for its potential benefits, whereas others fear that AI will hurt their privacy and ability to control their jobs, lives, and futures (Zhang and Dafoe, 2019), suggesting potential drawbacks of AI technology. This suggests that SMIs could trigger various perceptions among consumers that may lead to different outcomes. The challenge is to know the underlying psychological mechanisms to explain potential positive/negative outcomes, yet studies on the subject are rare but urgently needed. This dissertation investigates this phenomenon, specifically, potential benefits and drawbacks of using SMIs compared to (human) SMIs in social media advertising. Based on established theories of persuasion on advertising and brand endorsers, this dissertation identifies a crucial processing mechanism - source credibility - that is used to explain instances under which influencer type (i.e., AI vs. human) differentially influences advertising outcomes (i.e., attitude toward the advertisement, attitude toward brand, and purchase intentions). Source credibility is the source's truthfulness and believability perceived by the consumers (Roy et al., 2017). The treatment of source credibility as a mediating factor is a unique aspect of the research and that diverges from prior approaches that treat it as a predictor of persuasion. Rather, this research conceptualizes source credibility as dynamic, constantly changing, and not related in a simple way to the persuasiveness of an influencer type. Three dimensions of source credibility - expertise, trustworthiness, and physical attractiveness - are proposed to explain how an AI influencer may perform better/worse than a (human) SMI on advertising outcomes. A review of five decades' of source credibility studies noted several discrepancies regarding three dimensions of source credibility, suggesting gaps or unresolved issues in source credibility that deserve attention (see Pornpitakpan, 2004). Additionally, initial negative/positive dispositions toward the brand influencers, which has hardly been examined (see Pornpitakpan, 2004), will be measured in an effort to assess how much, if any, of a shift is detected in source credibility perceptions. To summarize, this research examines effectiveness of influencer type (artificial intelligence influencer vs. SMI) on persuasion of social media advertising (attitude toward the ad, attitude toward the brand, and purchase intentions). One mediator, source credibility, is proposed to explain the results. This is accomplished with a pilot study and an experimental study conducted in an online setting.