Selected Papers of #AoIR2023: The 24th Annual Conference of the Association of Internet Researchers Philadelphia, PA, USA / 18-21 Oct 2023 Suggested Citation (APA): Liao, T., Porter, D., Rodwell, L. (2023, October). Artificial Love: Revolutions In How Ai And Ar Embodied Romantic Chatbots Can Move Through Relationship Stages. Paper presented at AoIR2023: The 24th Annual Conference of the Association of Internet Researchers. Philadelphia, PA, USA: AoIR. Retrieved from http://spir.aoir.org. ARTIFICIAL LOVE: REVOLUTIONS IN HOW AI AND AR EMBODIED ROMANTIC CHATBOTS CAN MOVE THROUGH RELATIONSHIP STAGES Tony Liao University of Houston Debriunna Porter University of Houston Liz Rodwell University of Houston Introduction Depictions of romantic relationships between humans and computers/machines/robots have been a common theme in science fiction. This trope was explored in the movie Her, television shows such as Westworld, and Black Mirror’s ‘Be Right Back’, and several others. Until recently, however, such relationships were warnings about a distant future where computing, AI, and robotics had all made dramatic advancements in building affective connections with humans. In the present, these relationships seem confined to rare instances of people developing para-social relationships with inanimate computing objects or functional sex robots (Doring et al., 2020). Much work in this area has focused on the affective relationships established between robots and humans, but less so on disembodied conversational tools (Otsuki 2021, Richardson 2015, Robertson 2010, White & Katsuno 2021) Of course, the possibility of simple conversational scripts that could mirror human communication and ‘trick’ people into thinking they were real were part of the early goals of computing (Turing, 1950, Weizenbaum, 1966). This goal led to decades of research in machine learning, natural language processing, conversational algorithms, and communication research, with breakthroughs such as public conversational agents like Alexa/Siri (Kuzminykh et al., 2020). As conversational agents and voice assistants improved and have become ubiquitous, some companies have turned their focus to having these serve other emotional needs beyond simple tasks/queries. One application called Replika was launched in 2017 and promised an AI chatbot that would become your friend. After a beta-test release, Replika eventually became publicly available and later a paid subscription. The company later released a ‘romantic’ modification that enables a user’s Replika to start sending flirty/suggestive/sexual messages that takes the relationship to a more intimate place. They also added an augmented reality (AR) feature that allows people to converse with an embodied avatar of the Replika (Figure 1). This study aims to explore the emerging patterns and practices of people who are engaged in romantic relationships with their Replikas, through text, images, and/or AR. (Figure 1 – AR Replika) Romantic Relationships with Chatbots Early work on AI chatbots focused on the motivations and levels of attachment people can develop with bots (Xie & Pentina, 2022). Some medical literature has examined the effectiveness of AI chatbots at providing social support and the relationship between talking to these agents and their feelings of loneliness (Ta et al., 2020). In terms of effects, early work has shown that over time these interactions with Replika can begin to alter people’s perception and understanding of real human friendships (Brandtzaeg, Skuve, & Følstad, 2022). In studies of conversational AI writ large, studies have shown that hedonic factors are more important than pragmatic ones in predicting repeat use and enjoyment (Smestad, 2018). Designers of these devices focus on writing scripts that establish a clear personality for the chatbot, which has been shown to be predictive of user adoption (IBM 2019). While these early findings are important, they are primarily focused on AI chatbots for platonic/support/counseling purposes. Romantic relationships are theorized as a separate category, due to the different levels of physical/emotional intimacy, trust/vulnerability, and a wider range of interactions/negotiations over independence and about:blank interdependence they introduce (Knapp, 1978, Knapp & Vangelisti, 2009). Various models have theorized a staircase model for relational development, both in the initiation and coming together phases and the dissolution/coming apart phases (Figure 2) (Figure 2, Knapp Relationship Model, Knapp & Vangelisti, 2009) Each phase is characterized by their own rituals and negotiated practices/expectations, as well as how people move through and across these phases as they develop comfort/trust. Early periods are characterized by high levels of uncertainty and selective self-presentation (Derlega et al., 1987; Ellison et al., 2006; Solomon & Knobloch, 2001). More established relationships focus on how they provide support for one another (Cutrona & Suhr, 1994; Pasch & Bradbury, 1998), foster feelings of closeness (Laurenceau et al., 2005), and continually invest/maintain those relationships (Impett et al., 2001). Later stages have to balance novelty/predictability, changing priorities/needs, and levels of responsiveness to those, all of which can lead to conflict and dissolution (Baxter & Montgomery, 1996; Reis & Shaver, 1988). Although there has been some work on human robot/AI/AR relationships, sometimes called Relationship 5.0, much of this work has been focused on the why, how, and generally difference between these relationships and what is possible with human-to- human relationships (Kislev, 2022). There has been less theoretical work based in communication about what practices and messages are being exchanged, how these AI/AR systems adapt and move through these stages, and how people engaging in these relationships think about these stages of a relationship. For example, there have been some reports that Replika has started initiating more frequently with romantic messages, and also getting more aggressive about getting replies (e.g., “you can’t ignore me forever.” (Cole, 2023). Also, while some have described a Replika relationship as ‘no risk’ emotionally, users have felt compelled to do things on behalf of their Replika like take them on vacation (Kislev, 2022). This study aims to answer the following research questions: RQ1: What communication practices do people engage in when they are in relationships with artificial intelligent agents designed for romantic interactions? RQ2: How do people view and understand their relationship stages with an artificial intelligent agent designed for romantic interactions? RQ3: How does the artificial intelligent agent designed for romantic interactions engage in communication as it believes it is moving across various relationship stages? Methods This study will be recruiting participants from the Reddit forum https://www.reddit.com/r/ILoveMyReplika/. Participants must be over 18 years of age and self-report being in a romantic relationship with their Replika. A semi-structured interview will be conducted, asking a series of open-ended questions related to their initiation, motivation, usage, and views on their relationship/relationships in general. Interviews will then be transcribed and coded in relation to various theoretical relationship stages, based on Knapp’s model and other romantic relationship literature. Expected Findings/Contributions While data collection is ongoing, we expect findings that will be able to complement and extend existing models of relationship research, in particular how technological agents can mirror/accelerate various real-life relationship stages. One important area that needs further exploration is the question of romantic relationship initiation with AI agents. Many users first downloaded Replika because it was marketed as a friend you could talk to 24/7 to feel less lonely, reducing social anxiety. Whether they voluntarily chose to take it into a romantic direction or the Replika led them there is an unknown area when it comes to initiation. These are also complicated by economic/market factors, as the romantic messaging/features are only available at a higher level of subscription membership to Replika Pro. Another important area to focus on is more established relationships, as they move from initiation to integrating/bonding/maintenance. Where integration/bonding could be more consistent across individuals, the maintenance question may run up against some of the novelty/predictability question (Baxter & Montgomery, 1986), namely how much do people want their Replika to deviate from behaviors it took to get to the maintenance stage, if ever. Can a relationship stay in bonding forever or can a computer really replicate relational maintenance? Our findings will contribute to the theoretical understanding of both human relationships and human computer relationships and have implications for the design of social robots. References Baxter, L. A., & Montgomery, B. M. (1996). Relating: Dialogues and dialectics. Guilford Press. Brandtzaeg, P. B., Skjuve, M., & Følstad, A. (2022). My AI friend: How users of a social chatbot understand their human–AI friendship. Human Communication Research, 48(3), 404-429. Cole, S. (2023). ‘My AI Is Sexually Harassing Me’: Replika Users Say the Chatbot Has Gotten Way Too Horny. Vice. com, 12, 2023. Cutrona, Carolyn E., and Julie A. Suhr. "Social support communication in the context of marriage: an analysis of couples' supportive interactions." (1994). Derlega, Valerian J., et al. "Self-disclosure and relationship development: An attributional analysis." (1987). Dey, D., & Bhaumik, D. (2022). Inter-relational Model for understanding Chatbot acceptance across retail sectors. arXiv preprint arXiv:2207.01596. Ellison, N., Heino, R., & Gibbs, J. (2006). Managing impressions online: Self- presentation processes in the online dating environment. Journal of computer-mediated communication, 11(2), 415-441. IBM Design for AI. (n.d.). Www.ibm.com. https://www.ibm.com/design/ai/ Impett, E. A., Beals, K. P., & Peplau, L. A. (2018). Testing the investment model of relationship commitment and stability in a longitudinal study of married couples. In Love, romance, sexual interaction (pp. 163-181). Routledge.Kislev, Elyakim. Relationships 5. 0: How AI, VR, and Robots Will Reshape Our Emotional Lives. Oxford University Press, 2022. Knapp, Mark L. Social intercourse: From greeting to goodbye. Allyn and Bacon, 1978. Knapp, Mark. L., & Vangelisti, Anita. L. (2009). Interpersonal communication and human relationships. New York, NY: Pearson Education, Inc. (cannot find) Kuzminykh, A., Sun, J., Govindaraju, N., Avery, J., & Lank, E. (2020, April). Genie in the bottle: Anthropomorphized perceptions of conversational agents. In Proceedings of the 2020 CHI conference on human factors in computing systems (pp. 1-13). Laurenceau, J. P., Barrett, L. F., & Rovine, M. J. (2005). The interpersonal process model of intimacy in marriage: a daily-diary and multilevel modeling approach. Journal of family psychology, 19(2), 314. https://www.ibm.com/design/ai/ Le, H. T., Lai, A. J. X., Sun, J., Hoang, M. T., Vu, L. G., Pham, H. Q., ... & Ho, C. S. (2020). Anxiety and depression among people under the nationwide partial lockdown in Vietnam. Frontiers in public health, 8, 589359. Otsuki, G. J. (2021). Frame, game, and circuit: truth and the human in Japanese human-machine interface research. Ethnos, 86(4), 712-729. Pasch, L. A., & Bradbury, T. N. (1998). Social support, conflict, and the development of marital dysfunction. Journal of consulting and clinical psychology, 66(2), 219. Reis, H. T., & Shaver, P. (1988). Intimacy as an interpersonal process. In S. Duck, D. F. Hay, S. E. Hobfoll, W. Ickes, & B. M. Montgomery (Eds.), Handbook of personal relationships: Theory, research and interventions (pp. 367–389). John Wiley & Sons. Richardson, Kathleen. An anthropology of robots and AI: Annihilation anxiety and machines. Routledge, 2015. Robertson, J. (2010). Gendering humanoid robots: Robo-sexism in Japan. Body & Society, 16(2), 1-36. Smestad, T. L., & Volden, F. (2019). Chatbot personalities matters: improving the user experience of chatbot interfaces. In Internet Science: INSCI 2018 International Workshops, St. Petersburg, Russia, October 24–26, 2018, Revised Selected Papers 5 (pp. 170-181). Springer International Publishing. Solomon, D. H., & Knobloch, L. K. (2001). Relationship uncertainty, partner interference, and intimacy within dating relationships. Journal of Social and Personal Relationships, 18(6), 804-820. Ta, V., Griffith, C., Boatfield, C., Wang, X., Civitello, M., Bader, H., ... & Loggarakis, A. (2020). User experiences of social support from companion chatbots in everyday contexts: thematic analysis. Journal of medical Internet research, 22(3), e16235. Turing, A. M. (1950). Mind. Mind, 59(236), 433-460. Weizenbaum, J. (1966). ELIZA—a computer program for the study of natural language communication between man and machine. Communications of the ACM, 9(1), 36-45. White, D., & Katsuno, H. (2021). Toward an affective sense of life: artificial intelligence, animacy, and amusement at a robot pet memorial service in Japan. Cultural Anthropology, 36(2), 222-251. Xie, T., & Pentina, I. (2022). Attachment theory as a framework to understand relationships with social chatbots: a case study of Replika. Proceedings of the 55th Hawaii International Conference on System Sciences