Link to my Pinterest Board
The topic I've chosen is AI composition. This is an extension of AI narrative—musical expression is also a form of linear storytelling and emotion, playing a role similar to textual narratives.
AI composition is evidently not breaking news—I've come across literature discussing the application of related technologies over two decades ago. However, in recent years, with the gradual maturation of AI technology and breakthroughs achieved in the last couple of years, an increasing number of AI-generated musical pieces have emerged, accompanied by the rise of some new trends. For instance, AI mimicking the vocal tones of certain singers to sing songs by other artists has gained popularity. This breakthrough allows numerous singers, including some who have passed away, to "perform" songs that were otherwise beyond their reach. Such forms have become particularly trendy on platforms like YouTube, Twitter, and others in recent times.
The first artile delve into the world of AI-generated music, exploring its evolution and transformative impact on the creative landscape. It started by recounting the author's experience in music creation, highlighting the transition from pattern matching to AI algorithms. It spotlighted notable cases, like Taryn Southern's AI-composed album and the use of AI to recreate the voices of deceased artists. These examples raise legal and ethical considerations, showcasing both the potential and challenges of AI in music composition. The article concluded by showcasing two AI-generated songs and accompanying visuals created using MusicLM, DeepAI, and DALL·E 2[1].
This showcases a form of synesthesia—utilizing AI as a bridge to connect AI composition and AI painting. It analyzes how they express a shared emotion, exploring how they break through formal consistency on the level of "performance."
The second study emphasizes that Computational Thinking (CT) involves the problem-solving habits employed by computer scientists and programmers. The research contributes significantly to understanding how CT can facilitate interdisciplinary learning outcomes in both music and programming, specifically with Sonic Pi. The study identifies synergies, overlaps, and limitations in integrating CT concepts, practices, and perspectives, offering potential applications in New Zealand (NZ) and beyond within the Digital Technologies Curriculum (DTC)[2].
This study references a concept that is quite intriguing: abstraction and modularization, referred to as "building something large by assembling smaller parts together." By closely examining all the final code of each musical piece, the live looping Sonic Pi syntax is almost entirely used as code blocks for abstract music sequences, forming a modular structure. Live looping enables students to layer and repeat musical sound sequences in a hierarchical manner.
This kind of thinking is reminiscent of recursion and also brings to mind literature I read while studying AI narratives—integrating human feedback learning with recursive task decomposition: using models trained on smaller parts of the task to assist humans in providing feedback for a broader task. Gathering extensive demonstrations and comparisons from human labelers, GPT-3 is fine-tuned using behavior cloning and reward models to recursively summarize[3].
Finally, I would like to bring the timeline back to the present and touch upon some of the latest trends in the field of social media, along with feedback on legal and ethical aspects.
The third article focuses on a recent legal decision posed by YouTube, where thousands of content creators published "their" works of AI generated songs, usually "sung" by a popular artist to draw public attention. YouTube plans to implement two sets of content guidelines for AI-generated deepfakes, distinguishing between strict rules for music industry partners and looser rules for others. Creators will be required to label "realistic" AI-generated content, especially for topics like elections or conflicts. Penalties for inaccurate labeling may include takedowns and demonetization. YouTube will allow users to request removal of videos that simulate identifiable individuals. However, exceptions won't apply to AI-generated music content from YouTube's partners, potentially leading to takedowns for music covers. The platform aims to navigate the complex landscape of AI-generated content and its impact on various industries[4].
Apart from reminding me of the authorship debate in AI creation, fundamentally, it highlights the absence in AI storytelling. The prohibition of AI cover song videos on YouTube fundamentally reflects the challenges and dilemmas faced by AI in the realm of music. This may stem from concerns about AI-generated content, including potential copyright infringement, legal disputes, and the risk of spreading misleading information. It also highlights that social media platforms may adopt a relatively conservative stance when dealing with emerging technologies and innovative forms, as they strive to balance user creative expression with the responsibility to protect intellectual property.
From a storytelling perspective, this decision could exacerbate the cognitive gap regarding AI's role as a creator in the music domain, hindering the dissemination of the stories behind AI music composition. It suggests that the potential stories, creativity, and innovation of AI in the field of music may not receive sufficient recognition. The decisions made by social media platforms may influence public perceptions of AI's role and potential in music creation, raising questions about the legitimacy and artistic value of AI.
On a more fundamental level, this reminds me of a viewpoint from an article in Module 3: AI stories should inspire and promote scenarios beyond the existing power structures, rather than repetitively emphasizing and mimicking current narratives[5]. Presently, whether in AI avatars or AI music, both rely on viral social media dissemination. In my view, this essentially represents another reproduction of clickbait, tittytainment and mundane aspects of life. While the technological breakthrough is remarkable, the creativity in storytelling and meaning appears uninteresting, conventional, and imitative.
This phenomenon not only highlights a gap in the potential of AI stories to transcend current power structures but also underscores a missed opportunity for genuine innovation and creativity. The dominance of repetitive and imitative narratives on social media platforms may inadvertently stifle the exploration of alternative perspectives, novel ideas, and unconventional storytelling formats that AI could bring to the forefront.
In conclusion, the imperative lies in reimagining the role of AI stories as not just technological marvels but as agents of genuine creative disruption. Embracing diversity in narratives, encouraging experimentation, and challenging the status quo can propel AI-generated content into uncharted territories, fostering a richer, more dynamic storytelling landscape that transcends the limitations of current social media trends.
References:
[1] Freeman, J. (2023, July 12). AI in Music: A New Era of Composition. Harmonizing Technology and Creativity in AI Music Composition. Retrieved from https://jessefreeman.com/ai-in-music-a-new-era-of-composition-caaba6f00718
[2] Petrie, C. (2022). Interdisciplinary computational thinking with music and programming: a case study on algorithmic music composition with Sonic Pi. Computer Science Education, 32(2), 260-282. https://doi-org.login.ezproxy.library.ualberta.ca/10.1080/08993408.2021.1935603
[3] Wu, J., Ouyang, L., Ziegler, D. M., Stiennon, N., Lowe, R., Leike, J., & Christiano, P. (2021). Recursively Summarizing Books with Human Feedback. arXiv preprint arXiv:2109.10862. https://doi.org/10.48550/arXiv.2109.10862
[4] Sato, M., & Patel, N. (2023, November 14). YouTube is going to start cracking down on AI clones of musicians. The Verge. https://www.theverge.com/2023/11/14/23959658/google-youtube-generative-ai-labels-music-copyright
[5] Chubb, J., Reed, D., & Cowling, P. (2022). Expert views about missing AI narratives: is there an AI story crisis? AI & Society. https://link-springer-com.login.ezproxy.library.ualberta.ca/article/10.1007/s00146-022-01548-2
Eryi Your blog post offers a comprehensive and insightful analysis of the evolving landscape of AI composition and its implications within the realms of music, storytelling, and social media. You adeptly weave together various strands of research and contemporary developments to highlight the multifaceted impact of AI on creative expression and the accompanying ethical, legal, and societal challenges.
ReplyDeleteThe exploration of AI-generated music as both a continuation and an evolution of AI narratives is particularly compelling. By framing musical composition as a form of storytelling, you underscore the emotional and expressive depths that AI can reach, paralleling the complexity and nuance typically associated with textual narratives. This perspective not only broadens the understanding of AI's capabilities but also invites readers to reconsider the essence of creativity and authorship in the digital age.
The discussion on the role of Computational Thinking (CT) in bridging music and programming through algorithmic composition further enriches the narrative. By highlighting the educational and interdisciplinary benefits of CT, you reveal the potential for AI and technology to enhance creative processes in ways that are accessible, innovative, and inclusive. This segment of your analysis underscores the transformative power of AI in democratizing creative expression and fostering a deeper engagement with technology across diverse fields.
Reflecting on Module 5 readings: Thompson and Reilly (2019) explore the democratization of curation in the digital era, positing that with the rise of internet platforms, virtually everyone has become a curator. This broadening of the curatorial role reflects a significant shift in how we interact with and disseminate information and cultural content. The authors argue that this trend empowers individuals by allowing them to shape their own cultural landscapes, challenging traditional gatekeepers of culture and knowledge.
ReplyDeleteMihailidis and Cohen (2013) delve deeper into the implications of this shift for digital and media literacy education. They propose that curation should be considered a core competency in media literacy frameworks. By fostering critical thinking and selection skills, curation equips individuals with the ability to manage the overwhelming flow of digital information effectively. This approach not only enhances personal knowledge management but also encourages a more conscientious consumption and sharing of information.
Beth Anne Schwamberger's 2020 article on Pinterest strategies offers practical insights into how digital platforms facilitate and amplify the curatorial skills of their users. By leveraging algorithms and user engagement, platforms like Pinterest allow users to curate content in a way that not only reflects personal tastes but also engages broader audiences. This demonstrates how digital tools can enhance the visibility and impact of curated collections, making personal expression and influence more accessible than ever.
The Hermitage Museum's 2012 open letter provides a critical perspective on the dilution of the term 'curate' in popular culture. It emphasizes the depth of expertise and the deliberate methodology involved in traditional curation, underscoring a tension between professional curatorial practice and its colloquial use online. This critique invites reflection on the balance between the democratization of curation and the maintenance of rigorous standards that professional curators uphold.
Together, these sources paint a comprehensive picture of how digital platforms are reshaping the concept of curation. They highlight the transformative power of these tools in expanding the role of curator from a select few to a general populace, while also stressing the need for a nuanced understanding of what it means to curate in a world where information is abundant and diverse. This analysis underscores the importance of curation as a critical skill in navigating and shaping the digital landscape, suggesting its pivotal role in digital and media literacy education.