Artificial intelligence (AI) has found widespread application across various domains, including the arts. Generative AI, as a subfield of AI, is capable of assimilating artists’ styles, voices, and techniques, utilizing these components to generate novel artworks. A collaborative effort involving data scientists, developers, engineers, and art historians led to the creation of a new Rembrandt-style painting after an exhaustive examination of the entire collection of Rembrandt’s paintings (Blakemore, 2016). Tencent Music Entertainment, a Chinese entertainment company, developed and released 1,000 tracks featuring AI-generated vocals, one of which amassed over 100 million streams (Stassen, 2022). Notably, generative AI has been employed to finalize unfinished artworks initiated by deceased artists. For instance, in 2019, Huawei, a Chinese technology company, utilized AI through the Huawei Mate 20 Pro smartphone to complete Schubert’s Symphony No. 8 (Kennedy, 2019). Similarly, the BeethovenX AI project, comprising a group of data scientists and musicologists, completed Beethoven’s Symphony No. 10 with AI assistance (Caldwell, 2021). More recently, Paul McCartney, a member of The Beatles, employed artificial intelligence to create a new Beatles song using John Lennon’s voice (Pareles, 2023).

Current studies indicate a prevailing negative perception among consumers toward AI-generated hedonic products, attributing less value to them compared to those crafted by humans (Millet et al., 2023; Zhang et al., 2022). These investigations predominantly focus on artworks solely created by AI, positioning AI as a substitute for human creators. However, with the advancement of our comprehension of AI, it becomes evident that AI serves as a facilitator of human capabilities (De Cremer et al. 2021). It's imperative to acknowledge that the creation of all AI-generated artworks involves inputs from both human and AI. This holds particularly true in the context of AI-completed posthumous artwork. The creation of posthumous artworks represents a unique form of collaboration between deceased human artists and AI. Remarkably, previous research has yet to acknowledge this collaborative aspect. Moreover, current research primarily focuses on customers’ perceptions of the artwork itself, overlooking the influences of the AI-completed posthumous artwork on the deceased artists and their legacy.

This study aims to address these gaps by considering the AI-completed artworks as products emerging from the collaboration between humans and AI and examining consumers’ perceptions of AI-completed posthumous artworks and their impact on the legacy of the deceased artist. Our research question is “How does the human-AI collaborated completion of posthumous artwork influence consumers’ perception of the artworks and the artists’ legacy?” Answering this question should advance our understanding of consumers’ reactions to human-AI collaboration in the completion of posthumous hedonic products and thus help artists manage their personal brands and their legacy after they pass away.

We plan to adopt a mixed method in this study. A netnography has been done to explore consumers’ perceptions of AI-completed posthumous artworks. Following that, we proposed the conceptual model. The next step is to use experiments to examine the conceptual model. We aim to complete the experiment 1 early next year and the experiment 2 in summer 2024.

Exploratory Study and Result

As introduced above, we conducted a netnography as an exploratory study. Three specific cases were selected based on the degree of AI involvement in the completion of the artworks: the completion of Beethoven’s Symphony No. 10, wherein AI was predominantly responsible for completing the majority of the symphony due to the fragmented nature of Beethoven’s original musical sketches; the completion of Schubert’s Symphony No. 8 by Huawei with AI assistance, where AI contributed to composing the last two movements to complete the symphony; and The Beatles' new song 'Now and then,' created with AI support to incorporate the voice of the deceased artist, John Lennon where human artists completed majority of the creation work. Consumers’ comments on YouTube videos featuring these AI-completed posthumous artworks were collected and analyzed. Table 1 presents the information regarding the YouTube videos and associated comments included in this study.

--------------------------Insert Table 1 Here---------------------------

Upon analyzing numerous consumer comments, divergent attitudes among consumers regarding AI-completed posthumous artworks emerged in both the Beethoven and Schubert cases. Some individuals exhibited enthusiasm and satisfaction with the 'realization' of the unfinished artwork, while others expressed criticism regarding the ethicality of AI usage in art creation and the quality of the AI-completed music. Notably, among those holding positive attitudes toward the AI-completed artworks, many displayed favorable perceptions of AI. Conversely, individuals with negative attitudes toward the AI-completed artworks expressed concerns about the perceived authenticity of the artwork, ethical considerations, and apprehensions regarding AI potentially replacing human creative endeavors.

Interestingly, we did not observe a divergence in attitudes in the case of The Beatles. All comments exhibited a supportive stance toward both the artwork and the artists, resonating with a pervasive sense of nostalgia evident throughout the commentary. This unanimity might be attributed to the relatively limited level of AI involvement in this instance. Additionally, it could be linked to consumers’ emotional attachment to John Lennon. Another potential factor could be the genre disparity: while the Beethoven and Schubert symphonies belonged to the high-brow classical music category, The Beatles' song falls under the low-brow pop music genre.

Conceptual Framework and Future Experiments

Based on the findings of the exploratory study, our proposition suggests that the extent of AI involvement in completing the posthumous artwork significantly affects consumers' perceived authenticity of the artwork and their perceived ethicality regarding AI utilizaiton in this context. These perceptions, in turn, influence consumers’ attitudes toward both the artwork and the associated artists. Additionally, consumers' attitudes toward AI and their connection with the artist serve as moderators in this relationship. Figure 1 illustrates the conceptual model.

-------------------------Insert Figure 1 Here----------------------------

Planned Experiment 1

To examine the conceptual model, we intend to conduct an experiment manipulating the degree of AI involvement in the posthumous artwork completion. M-Turk will be used to recruit participants. Experiment participants will be randomly assigned to one of the four conditions: AI completed the majority of the creation work, completed half of the creation work, completed a limited percentage of the creation work, and a control group. Each participant will receive a one-page description illustrating how AI is employed to complete the unfinished artwork of a fictitious deceased artist. Subsequently, we will assess participants’ attitudes toward AI, perceived authenticity of the artwork, perceived ethicality of AI usage, attitude toward the artwork, and attitude toward the artist.

Planned Experiment 2

Given the observed negative attitudes toward AI-completed artwork in the case of Beethoven and Schubert, we also intend to investigate the mechanism that mitigates the harm of using AI to complete posthumous artworks. We hypothesize that if the family of the deceased artist or professionals endorse the utilization of AI in completing the posthumous artwork, consumers’ perceived legitimacy will increase, consequently reducing negative perceptions. To test this hypothesis, we plan to conduct an online experiment employing a 4 (legitimized by family, legitimized by professional, legitimized by an arts organization, legitimized by a commercial organization) * 4 (AI completed the majority of the creation work, AI completed half of the creation work, AI completed a limited percentage of the creation work, and the control group) design. Experiment participants recruited from Amazon Mechanical Turk (M-Turk) will be randomly assigned to one of the eight conditions. We will measure consumers’ attitude toward AI, their perceived legitimacy of using AI to complete the posthumous artwork, perceived authenticity of the artwork, perceived ethicality of AI usage, attitude toward the artwork and attitude toward the artist.

Contribution And Takeaway

From a theoretical standpoint, this study advances our comprehension of AI-generated hedonic products by transitioning from an AI solo creation viewpoint to a focus on human-AI collaboration. AI is viewed as a creation facilitator rather than a substitute for the human creative process. Furthermore, this study contributes to the literature on human-AI collaboration. The level of AI involvement in the product creation process significantly influences consumers’ responses to the creative outcomes.

Reference

Blakemore, Erin (2016, April 5), “ ‘New’ Rembrandt Created, 347 Years After the Dutch Master’s Death”, Smithsonian magazine, https://www.smithsonianmag.com/smart-news/new-rembrandt-created-347-years-after-the-dutch-masters-death-180958664/
Caldwell, Vanessa (2021, November 22), “Beethoven started composing his 10th Symphony in the 1820s. AI finished it in 2021”, CBC, https://www.cbc.ca/documentaries/the-nature-of-things/beethoven-started-composing-his-10th-symphony-in-the-1820s-ai-finished-it-in-2021-1.6258415
De Cremer, David and Kasparov, Garry (2021), “AI should augment human intelligence, not replace it”, Harvard Business Review, https://hbr.org/2021/03/ai-should-augment-human-intelligence-not-replace-it
Pareless, Jon (2023, November 2) “The Beatles’ ‘Now and Then’: A Glimpse of Past Greatness”, The New York Times, https://www.nytimes.com/2023/11/02/arts/music/beatles-now-and-then-last-song.html
Millet, K., Buehler, F., Du, G., & Kokkoris, M. D. (2023). Defending humankind: Anthropocentric bias in the appreciation of AI art. Computers in Human Behavior, 143, 107707. https://doi.org/10.1016/j.chb.2023.107707
Zhang, H., Bai, X., & Ma, Z. (2022). Consumer reactions to AI design: Exploring consumer willingness to pay for AI‐designed products. Psychology & Marketing, 39(11), 2171–2183. https://doi.org/10.1002/mar.21721
Stassen, Murray (2022, November 15), “Over 1,000 songs with human-mimicking AI vocals have been released by Tencent music in China. One of them has 100M streams.” Music Business Worldwide, https://www.musicbusinessworldwide.com/over-1000-songs-human-mimicking-ai-vocals-have-been-released-by-tencent-music-in-china-one-of-them-has-over-100m-streams/
Kennedy, John (2019, February 22), “How AI completed Schubert’s Unfinished Symphony No 8”, Silicon Republic, https://www.siliconrepublic.com/machines/unfinished-symphony-no-8-ai-huawei

Museums and cultural heritage have financially relied on traditional forms of support, such as public subsidies, philanthropy, sponsorships and patronage. This paper empirically investigates current best practices and possible future developments in terms of business model innovation and applied novel digital technologies. The main addressed question is what the future of museum financing could look like, and corresponding implications. First, the museum and cultural heritage fields are gradually evolving into a sector where experimentation in areas such as financing activities through licensing income, crowdfunding, and blockchain technology is beginning to emerge. Second, digitalisation and changes in access to traditional sources of funding may in addition pave the way for business model innovation to optimise both the cost structure and organisational configurations of these arts and cultural institutions. The paper discusses: what kinds of changes can be observed; why they are observed; the efficiency and efficacy of some of the novel models in the context of the museum and cultural heritage sectors using empirical examples. Empirical examples of current and future developments related to combining funding and financing methods, as well as the adoption of novel business models, are used to answer the proposed questions.