1. Motivation and background
Henrik Hagtvedt and Vanessa M. Patrick (2008) introduced the art infusion effect into literature 15 years ago and confirmed it empirically. In essence, the art infusion effect states that the visible association of a commercial object (e.g., product) with art automatically leads to a more positive evaluation (attitude, purchase intention, etc.) of the object. The prerequisite for this positive spillover effect is that consumers perceive the art as art. Interestingly, this positive spillover effect of art on object evaluation occurs for both positive (Art Infusion Effect I) and negatively evaluated art (Art Infusion Effect 2). The reason for this effect is that art itself has a value that evokes luxury associations such as uniqueness, prestige, and culture (Lee et al. 2015; Hagtvedt & Patrick 2008), which are then transferred to the object.
This effect has been investigated and confirmed in various studies over the last 15 years (for overviews see Gupta & Joshi 2023; Baumgarth & Wieker 2020; Baumgarth 2018). Various determinants such as object characteristics (e.g., stores and retailing: Logkizidou et al. 2019; Naletelich & Paswan 2018), personal characteristics (e.g., regulatory fit: Mantovani & Tazima 2016; value-expression: Quach et al. 2022) and art characteristics (e.g., mentioning the artist's name: Marin et al. 2022) have also been analyzed as amplifiers and limits of the art infusion effect.
One of the most important developments in art infusion research was the expansion of the art considered. While Hagvedt and Patrick (2008) and many other researchers have considered classical and predominantly iconic pieces of visual art (e.g., Estes et al. 2018), in recent years there have also been an increasing number of studies that consider more diverse genres and types of art (e.g., cultural-related art: Seo et al. 2021; urban art: Baumgarth & Wieker 2020; contemporary art: Lee et al. 2015; abstract vs. realistic art: Naletelich & Paswan 2018; music: Cuny et al. 2020).
However, the (art) world does not stand still and constantly evolves. One of the latest art trends is AI art. At the latest since the Portrait of Edmund de Belamy in 2018, the art world has been discussing AI as an independent artist. This discussion has been further intensified by the emergence of generative AI and the availability of tools such as ChatGPT, Midjoureny, and DALL-E 2 for all. To date, there have been no investigations into whether and how the Art Infusion effect also applies to AI art. The paper therefore addresses the following research questions:
1. Is AI art (with or without a human artist) even perceived as art by consumers and valued in a similar way to "classical" art?
2. Does AI art create the art infusion effect in a similar way to human art?
3. Does this AI art infusion effect depend on whether a human artist is involved in the creation of the art (in-the-loop) or not (out-the-loop)?
2. AI art, art perception and AI art infusion effect
AI art refers to artworks created either by hybrid human-AI teams (in-the-loop), for example Refik Anadol and AI exhibitions such as “Unsupervised” (2023) or AI creating own artistic outputs (out-the-loop) based only on initial training data, for instance the “Portrait of Edmund de Belamy” sold by Christie’s for $432,500 (Christies 2018). AI art challenges traditional notions of authorship and copyright as well as raising questions about the nature of creativity and whether AI art is considered equally as creative as human art (Ajani 2022; Kalpokiene & Kalpokas 2023).
Literature assessing perceptions of AI versus human art find that when the same artwork is disclosed as human, it is more considered “art” then when disclosed as AI (Hong 2018). In various experiments, paintings disclosed as AI-created are liked significantly less, perceived as not as novel, meaningful, or beautiful than those disclosed as human created (Ragot et al. 2020). Further research shows a prevalent bias against AI-generated artworks and provides insights into the psychological factors driving this bias. Identical artwork receives lower preference when attributed to AI creation rather than human creation, primarily due to the perception of AI-made art as less creative, resulting in a diminished experience of awe, an emotional response commonly linked to the aesthetic appreciation of art (Millet et al. 2023). This inclination toward a negative bias of AI-disclosed artwork suggests that awareness of human involvement in the artistic process positively influences evaluations of art and this positivity stems from the significant value we place on the infusion of the human experience into art (Bellaiche et al. 2023). In summation, AI art is perceived as art by consumers, but not as strong as classical art. Thus:
H1: AI art is perceived as art in a similar way to human art.
In addition, a distinction must be made as to whether the art is created entirely by AI (out-the-loop) or whether the human artist creates the artwork together with the AI (in the loop). Research in the field of poetry shows that the perceived quality of the poem can be increased through AI and humans collaboration (Köbis & Mossink 2021). Furthermore, a study shows that the perceived authenticity of music increases when AI does not create it alone, but in collaboration with humans (Jago 2021). Thus:
H2: AI art is perceived more strongly as art when a human is involved as an artist (in-the-loop) than when the AI acts autonomously (out-the-loop).
Based on the classic art infusion effect and the research outlined above on the perception of AI as art and, in particular, AI + humans as art, the following hypotheses can be derived:
H3: Similar to human art, AI art in product and brand design has a positive influence on a) brand attitude and b) purchase intention.
H4: AI art with human involvement (in-the-loop) in product design has a stronger positive influence on a) brand attitude and b) purchase intention than AI art without human involvement (out-the-loop).
3. Studies
To test the four hypotheses, a two-step approach is chosen. In a first study, the two hypotheses on the perception of AI art are tested. Based on the design of Ragot et al. (2020), three AI artworks and three human artworks are selected. These are additionally disclosed with a label and a short story about their creation as AI, human or AI & human artwork, whereby this disclosure is correct in some cases and not in others. Following each image, six groups of consumers (Prolific sample) rate the extent to which the visual objects are art on a seven-point scale (Hagtvedt & Patrick 2008). In addition, the subjects also rated the images in terms of classic characteristics such as liking, beauty, profundity and worth (Bellaiche et al. 2023; Augustin et al. 2012), which are regularly used to evaluate visual artworks. When selecting the artworks, care is taken to ensure that the AI and real images are similar (one portrait, one still life and one abstract image), have the same format and that few are known. The databases Wikiart (wikiart.org), Europeana (www.europeana.eu) and AI Art Collection (https://aiartcollection.com/) serve as the basis for the image selection. Study 1 is to be carried out at the end of 2023.
In the second study, based on the study by Baumgarth & Wieker (2020), several packages will be designed for the two product categories of beer and household cleaners which, based on the results of the first study, will feature a work of art on the packaging labeled as AI, Human or AI & Human art. In addition, a control group is formed, which evaluates the fictitious beer or household cleaner brand without art integration. Brand attitude (Hagtvedt & Patrick 2008) and purchase intention (Newman et al. 2018) are measured using established scales. Also, perceived art is surveyed as a manipulation check, as in study 1. In addition to demographic questions, further control questions on interest in art (Hüttl-Maack 2018) and general AI attitudes (Grassini 2023) are also asked. The experiment will be conducted as a paper&pencil survey with a student sample in order to control the equality of the stimulus presentation. Study 2 is to be conducted in spring 2024.
4. Preliminary conclusions and next steps
The art infusion effect is a robust phenomenon that explains the effects of art integration on product and brand evaluation and thus represents an important basis for art-company collaborations. At the same time, the world and the art world are currently changing due to the establishment of new technologies such as AI. AI is also becoming increasingly powerful in the creative-artistic environment, which is why there is an intensive discussion around the topic of AI art. The subject of this work-in-progress paper is to analyze whether the Art Infusion effect is also valid for AI art and whether this effect depends on the type of AI art (in- or out-the-loop). The results for both studies will be available by the AIMAC conference and it is planned to present them there for the first time.

Christian Holst and Olga Kolokytha

The digital transformation has significantly changed the way social contexts are created and organized, as well as the creation and consumption of creative products and services. These new advances have led to a blurring of the distinction between producer and recipient where co-creative forms of production become more important and ubiquitous (Nakajima, 2012; Shirky, 2008). However, cultural institutions are “poorly equipped to deal with such an inundation of new claims to meaning“ (Stalder, 2018, p. 4). Our contribution looks at how organizational adaptation to participatory forms in the cultural sector can be enacted. We build on a case study of Burgtheater Vienna, which hosted a series of co-creative theatre events on Twitter (now X) during the coronavirus lockdowns and showcase how digital media can impact arts management practices with an emphasis on artistic planning, production and project management.

Theoretical background and research context

We build on two theoretical perspectives – co-creation and enactment theory. The term co-creation is used to describe the creation of value with the help of external resources like customers, citizens, and other organizations (Payne et al., 2008; Prahalad & Ramaswamy, 2004a, 2004b). While the literature mainly focuses on the general nature of co-creation and its pros and cons in an organisational context (Brandsen & Honingh, 2015; Iglesias et al., 2020; Woratschek et al., 2020), we know little about how it actually comes about, especially in creative, artistic processes. We therefore complement the co-creation perspective with the enactment lens, as described by Weick (1988, 2003). This lens puts a focus on how “extracted cues” (Weick, 1995, S. 49), i.e. key stimuli in the organisation’s environment, influence the organization's interpretation and actions. These actions in turn shape and change the environment, which then again offers new stimuli etc.

An interactive, open medium like Twitter (now X) gives us the opportunity to observe this process in a condensed form. We therefore explore seven Twitter theatre events at the Burgtheater, taking place between May 2020 and April 2022. For these events, the audience was invited to use Twitter, to create a collective narrative of an evening at the theatre after some basic frame and prompts given by the theatre. The development of events was left to the audience who created the “plot” and content of the event. Our study investigates how the co-creative event was enacted and examines the role of the theatre as the host in the process.

Our research questions are:
● How was the co-creative event enacted and which elements and dynamics turned out to be crucial for this enactment?
● How did the theatre as the hosting organisation understand its role in the process of enactment?

Methodology

Methodologically, a mixed methods approach seemed appropriate. In the first step, we used network analysis (Donato et al., 2017; Luke, 2015) for overviewing and structuring nearly 10,000 tweets from around 1,100 contributors. Using R and Python software, we visualized a network of nodes representing individual accounts engaged in the event (see fig. 1). The network was constructed using density and three centrality measures: degree (total number of connections), closeness (number of actors with whom the node has connections), and betweenness (stands between two actors). In a second step, we conducted an AI-supported thematic analysis (Braun & Clarke, 2012, 2019) of the tweets of the seven events. Furthermore, we conducted and analysed four interviews with event producers at different stages in the process. These interviews provided insights into the theatre's role in shaping the co-creation experience and the conclusions the theatre drew from the experience for future planning and projects.

Findings

Based on the network analysis, three different roles within the network can be categorised. (1) The theatre as host has the central role as the hub of interaction. (2) There are also contributors of varying influence, interacting with the impulses from the theatre. (3) Finally, there are spectators whose interactions are limited to small groups unconnected from the intensive interaction “centerstage”. This structure turned out to be resistant to conceptual changes among the different events.

The analysis of tweets from virtual theatre-goers revealed that despite the digital format's potential for role fluidity, participants largely adhered to traditional theatre roles and norms. They engaged in social commentary about the experience, discussed audience behaviour, and shared personal emotions, but did not challenge the foundational premises set by the theatre. Members of the theatre also maintained their conventional roles. The theatre acted as both initiator and facilitator, reinforcing traditional elements even in a digital setting.

Interviews with the theatre revealed a shift towards stronger pre-structuring in their events. While initially more ad hoc, the format evolved to a series with a more pre-structured plot and coordinated responses from theatre personnel. However, the theatre aimed to maintain an experimental approach, avoiding the repetition of a once-successful model. The focus shifted from a rigid storyline to preparing various cues, allowing actors to coordinate and react spontaneously. This change was driven by the realization that the essence of theatre lies in the shared experience of play and community.

Discussion and conclusion

By combining co-creation and enactment theories, our research offers a more nuanced perspective on the micro-processes underlying co-creation initiatives. The findings from the network analysis reflect the spatial setup of traditional theatre settings with a centerstage, where the action takes place, and an offstage space for only small-scale networked commentary like a theatre foyer. However, the role of theatre itself has fundamentally changed in this setting: It does not work as a provider of the creative offer but as a facilitator of “play and community”.

Furthermore, our findings exemplify the concept of "extracted cues" as a catalytic converter for co-creative interplay among the theatre as host and the contributors within the network. Unlike traditional organizations that rely on specific coordination mechanisms (Mintzberg, 1993), the co-creative digital setting relies on conventions familiar to all participants from real-life theatre experiences. In this sense, our findings can be related to Becker's explanations that collaboration in the artistic field is governed to a significant extent by conventions (Becker, 1982, S. 40–67). But beyond that, our study indicates that conventions do not only set the boundaries and constraints for the events but mainly provide a “recognition grid” for extracted cues. This allows for playful handling of conventions and an understanding of the essence of theatre as “play and community”.

As for our question about the role of theatre, the theatre iteratively refined the event structure over time by working in the mode of circular experimentation. This aspect aligns with Weick's notion that enactment is a dynamic process that is dependent on cues from the environment. Constant experimentation can also be seen as a genuinely artistic approach. But the way it was done here also goes beyond the theatre's traditional planning processes. It shows an opportunity for ad hoc creativity that is not possible within the "black box" theatre but in direct exchange with external stakeholders already during the creative process.

References

Becker, H. S. (1982). Art Worlds. University of California Press.
Brandsen, T., & Honingh, M. (2015). Distinguishing Different Types of Coproduction: A Conceptual Analysis Based on the Classical Definitions. Public Administration Review, 76(3), 427–435. https://doi.org/10.1111/puar.12465
Braun, V., & Clarke, V. (2012). Thematic Analysis. In H. Cooper (Hrsg.), Handbook of Research Methods in Psychology, Vol. 2 (S. 57–71).
Braun, V., & Clarke, V. (2019). Reflecting on reflexive thematic analysis. Qualitative Research in Sport, Exercise and Health, 11(4), 589–597. https://doi.org/10.1080/2159676X.2019.1628806
Donato, H. C., Farina, M. C., Donaire, D., & Santos, I. C. D. (2017). Value Co-Creation and Social Network Analysis on a Network Engagement Platform. Revista de Administração Mackenzie, 18(5), 63–91. https://doi.org/10.1590/1678-69712017/administracao.v18n5p63-91
Iglesias, O., Markovic, S., Bagherzadeh, M., & Singh, J. J. (2020). Co-creation: A Key Link Between Corporate Social Responsibility, Customer Trust, and Customer Loyalty. Journal of Business Ethics, 163(1), 151–166. https://doi.org/10.1007/s10551-018-4015-y
Luke, D. A. (2015). A User’s Guide to Network Analysis in R. Springer. https://doi.org/10.1007/978-3-319-23883-8
Mintzberg, H. (1993). Structure in Fives. Designing Effective Organizations. Prentice-Hall.
Nakajima, S. (2012). Prosumption in Art. American Behavioral Scientist, 56(4), 550–569. https://doi.org/10.1177/0002764211429358
Payne, A. F., Storbacka, K., & Frow, P. (2008). Managing the co-creation of value. Journal of the Academy of Marketing Science, 36(1), 83–96. https://doi.org/10.1007/s11747-007-0070-0
Prahalad, C. K., & Ramaswamy, V. (2004a). Co‐creating unique value with customers. Strategy & Leadership, 32(3), 4–9. https://doi.org/10.1108/10878570410699249
Prahalad, C. K., & Ramaswamy, V. (2004b). Co-creation Experiences: The next practice in value creation. Journal of Interactive Marketing, 18(3), 5–14. https://doi.org/10.1002/dir.20015
Shirky, C. (2008). Here Comes Everybody: The Power of Organizing Without Organizations. Penguin Books. https://doi.org/10.3399/bjgp09x420437
Stalder, F. (2018). The Digital Condition. Polity Press.
Weick, K. E. (1988). Enacted Sensemaking in Crisis Situations. Journal of Management Studies, 25(4), 305–316.
Weick, K. E. (1995). Sensemaking in Organizations. Sage Publications.
Weick, K. E. (2003). Enacting an Environment: The Infrastructure of Organizing. In Debating Organization. Point-counterpoint in Organization Studies. Blackwell Publishing.
Woratschek, H., Horbel, C., & Popp, B. (2020). Determining customer satisfaction and loyalty from a value co-creation perspective. Service Industries Journal, 40(11–12), 777–799. https://doi.org/10.1080/02642069.2019.1606213