Listening Challenge: Human or AI?

• Let’s start with a quick listening challenge.
• You will hear 4-5 short music clips.
• One clip is part of an original song segment I created myself.
• The other clips are continuations generated by Suno AI, prompted using my original piece.
• Your task: Listen closely and decide which clip (A, B, C…) you think is the original human creation.

Abstract Overview

• Suno generates music via prompts, bypassing traditional composing/performing.
• Raises questions about authenticity without a conventional “auteur”.
• Paper argues: AI music can achieve authenticity, leading to a new (contested) form of musicianship.

Introduction: Before Suno

Early AI & Music Experiments:

• 1957: “Illiac Suite for String Quartet” (Hiller & Isaacson) - Early computer composition.
• 1965: Ray Kurzweil’s pattern-based composition software.
• 1997: David Cope’s EMI program mimics composer styles.
  • Tech Closeup: Music Professor - YouTube
  • Albums released using “Emily Howell” algorithm (2009, 2012).

Introduction: The Rise of Generative AI Music

• 2024: Platforms like Suno and Udio bring generative AI music to the masses.
• Challenge: Music’s complex structure (sound + time) made AI progress slower than in text/image.
• Suno’s Breakthrough: Text/audio/image/video-to-music interface aims to bridge this gap.

Introduction: Suno’s Impact & Debate

• Suno pushes boundaries of “performance” and “composition”.
• Invites listeners to become producers/creators.
• BUT: Lack of direct human performance sparks debate on authenticity and agency.
  • Is prompting truly active composition or closer to passive consumption?
  • The paper acknowledges this debate

Introduction: Key Research Questions

1. Are we all musicians now? Has Suno democratized musicianship?
2. Can AI-generated music be authenticated? Is there potential for it to be considered “real”?
3. How might AI impact music scenes? Sociocultural dynamics, roles, production futures.

(What does “authenticating music” mean in this context?)

Authenticity in Music: “Real” Music, “Real” Musicianship

• Authenticity = “real,” “genuine,” “honest,” “sincere.”
• Crucial concept in (ethno)musicology, especially popular music studies.
• Discourses of authenticity assign musical value.

Authenticity: Musical vs. Non-Musical Dimensions

• Musical Authenticity: Based on sound elements.
  • Examples: Swing feel in jazz, electric bağlama in arabesque, specific guitar tones in metal.
  • (Suno can replicate these sonic markers).
• Non-Musical Authenticity: Based on artist identity, origin, context.
  • Examples: Gangsta rap credibility (Taylor, 1997), Chicago blues tourism (Becker, 2004).
  • (Suno inherently lacks this biographical/contextual dimension).

Authenticity: Ascribed, Not Inscribed

• Key Idea (Moore, 2002): Authenticity is assigned by listeners/communities, not an inherent property of the music itself.
• Values vary based on individual/collective perspectives.
• No “universal” or “absolute” values defining authenticity for any genre.

Authenticity & AI Music Perception

• Because authenticity is ascribed, AI music could be perceived as “authentic” or “artificial” depending on the listener’s value system.
  • (Connects to cultural relativism: values are context-dependent).

Discussion Point: Ascribed Authenticity

• How does the idea that authenticity is ascribed (assigned by listeners) change how you think about AI-generated music?
• Could you personally accept AI music as “authentic” under certain conditions?
• What might those conditions be (e.g., based on sound quality, emotional impact, novel style, transparency about its creation)?

Defining the “Musician”

• Crucial for analyzing the role of AI music generator users.
• Question seems obvious, but consider edge cases:
  • Is a DJ a musician?
  • A conductor?
  • An amateur vs. a professional?
  • Someone singing chants at a football match?

Defining “Musician”: Turino’s Framework

• Presentational Performance: Clear artist/audience separation (e.g., concert).
• Participatory Performance: Blurred lines, focus on collective engagement (e.g., campfire singalong, religious service, football chants).
  • Musicianship is ambiguous from an emic (insider) perspective. Are participants “musicians”?

Defining “Musician”: A Working Definition

• Adopted definition (Sağıroğlu, 2019):

  Musicians = “subjects who intervene in sounds and silences accepted as music with a specific intention (performance or composition).”

Discussion Point: Applying the Definition

• Using Sağıroğlu’s definition (“intervening in sounds… with intention”), where do these roles fit?
  • A DJ mixing tracks?
  • A conductor leading an orchestra?
  • Someone using Suno with a detailed prompt?
  • Someone using Suno with a simple prompt (“happy pop song”)?
• Does this definition feel too broad, too narrow, or about right?

Suno: The AI Music Generator

• Timeline: Initial release (Dec 2023), v3 (Mar 2024), v3.5 (May 2024), v4 (Nov 2024).
• Founding Team: Background in machine learning (e.g., at Kensho, working on audio transcription).
• Core Goal: “building a future where anyone can make great music… No instrument needed, just imagination.” (Suno website)

Suno: Raison d’être & Capability

• Suno’s Claim: “From your mind to music.”
• v3: Aimed for “radio-quality” music (2 mins).
• v3.5: Extended duration (4 mins), improved song structures.
• v4: Better audio quality, sharper lyrics, dynamic structures.

Discussion Point: Creator or Commissioner?

• Suno aims to turn imagination (“your mind”) into music. How well does it deliver on this promise from your perspective or experience?
• When using a tool like Suno, does the user feel more like the creator of the song, or like someone commissioning a song from the AI?
• Where does the creative agency seem to lie?

Suno: User Interaction & Output

• Simple Mode: Generate based on genre/style/theme description (instrumental option available).
  • “Need ideas?” feature guides prompts (person, occasion, mood, activity).
• Custom Mode: More control (user provides lyrics, etc.).
• Output: Generates full song (vocals, instruments, lyrics, title, artwork) from prompts like “psychedelic UK garage song about a friend with a Nokia obsession.”

Suno: Ownership & Copyright

• Usage Rights: Vary by subscription (Free, Pro, Premier).
  • Pro/Premier: Can monetize generated music (YouTube, Spotify, etc.).
  • Free: Non-commercial use only.
• Controversy: AI models trained on vast datasets, raising copyright concerns.
  • Example: “Heart On My Sleeve” (Fake Drake/Weeknd track) taken down after UMG request.

Analyzing Suno’s Impact: Four Key Topics

1. Collective Musical Intelligence: AI’s collaborative nature?
2. Musicalization of Everyday Life: Unconventional music creation.
3. Embracing Creation Playlists: Beyond passive listening.
4. Is There an Author? Locating creativity behind the machine.

1. Collective Musical Intelligence: AI as “Other”?

• Common Perception: AI often viewed as an external “Other” (influenced by sci-fi like Terminator, Matrix).
• Distinction: Need to separate narrow AI (like Suno) from hypothetical AGI (Artificial General Intelligence).
• Suno’s Foundation: Built on anthropocentric training data (recorded music, speech). Founders acknowledge this but are vague on specifics.

1. Collective Musical Intelligence: The Concept

• Argument: Generative AI represents a form of human collective intelligence, accessed via prompts.
• Musical Context: This can be termed “collective musical intelligence.”
• Suno as Access Point: Makes musical heritage (within training data limits) available to users, regardless of their prior musical knowledge (cultural capital).

1. Collective Musical Intelligence: Authenticity & Challenges

• Potential for Authenticity: Viewing AI output as accessing “collective musical intelligence” allows seeing the “human” behind the “machine,” potentially grounding authenticity. (Analogous to Anderson’s “imagined communities”).
• Challenges:
  • Lacks embodied cultural capital (learned skills, lived experience).
  • Relies on prompts + recorded data, not traditional musicianship.
  • Lack of transparency about training data fuels copyright issues & perceptions of inauthenticity.

Discussion Point: Collective Musical Intelligence

• Does framing AI music generation as accessing “collective musical intelligence” resonate with you?
• Does this perspective make AI music feel potentially more “authentic” or meaningful?
• Or do the challenges (lack of embodiment, transparency issues) outweigh this conceptual reframing for you? Why?

1. Collective Musical Intelligence: Social Construction

• Avoid technological determinism (idea that tech dictates social change).
• Technology adoption is shaped by social negotiation.
• Future: Legal/ethical debates around training data and ownership will shape the fate of AI music generators and their influence on music scenes.

2. Musicalization of Everyday Life

• Traditional Music Production: Labor-intensive, requires capital (cultural, social, economic), time.
• Suno: Technology-intensive, compresses process into seconds via prompts.
• Potential: Facilitates experimental/unconventional music; enables mass participation in the musicalization of everyday life.

Discussion Point: Musicalizing Daily Life

• What are your thoughts on AI enabling the “musicalization of everyday life”? Is this an exciting prospect or a potential source of noise/trivialization?
• What are the potential benefits (e.g., personal expression, fun, accessibility)?
• What are the potential drawbacks (e.g., devaluing crafted music, information overload, lack of intention)?

4. Is There an Author? Behind the Machine

• Author: Perceived creative agent behind a work.
• Complexity: Clearer in Western art music (composer) vs. popular music (multiple contributors: producer, songwriter, artist, etc.).
• How does this apply to Suno? Three perspectives…

4. Authorship Perspective

1: The Void / Simulacra

• Music lacks a “real” author: no “real” voices, instruments, or direct human performance (except potential audio uploads).
• Hiatt (2024) on Suno blues: “no human behind the voice, no hand on the guitar.”
• Baudrillard’s Simulacra: Representations detached from reality, forming their own hyperreality. Suno produces musical simulacra, severing the link to original human actions/intent.

4. Authorship Perspective

2: The Collective Subject

• Examine the training data (the “collective musical intelligence”).
• Authorship resides not in one person, but in the collective of musicians/creators whose work informed the AI.
• Challenges traditional notions but recognizes authorship can evolve.

4. Authorship Perspective

3: The User as Author/Curator

• Users are responsible for initiating unique productions via prompts (detailed or simple).
• Even simple prompts involve curation (selecting the prompt itself).
• Curation Actions: Extending, cropping, replacing sections = creative filtering (“selecting and discarding”).

Discussion Point: Locating Authorship

• Which of the three perspectives on AI music authorship resonates most with you?
  • The Void / Simulacra (No real author)
  • The Collective Subject (Authorship = the training data contributors)
  • The User as Author/Curator (Authorship = the prompter/editor)
• Can these perspectives coexist? For instance, could the user be the curator while the collective provides the raw material?

Discussion: Are We All Musicians Now?

• Answer: Potentially “Yes” AND “No.”
• Using Sağıroğlu’s Definition: (“intervene in sounds… with specific intention”)
  • Suno users do intervene: selecting genre, mood, theme, lyrics, instruments, uploading audio.
  • They act with intention (to compose/create a piece of music).
• Therefore, users could be positioned as authors and musicians.

Discussion: Musicianship as Intervention & Curation

• Intervention: Users make choices about genre, style, mood, theme, lyrics, instrumentation.
• Curation:
  • Selecting specific outputs from multiple generations.
  • Editing: “extending,” “cropping,” “replacing sections.”
• This “selecting and discarding” is akin to DJ curation and can be seen as musicianship.
• Amateur vs. Professional: Tied to subscription plan/monetization rights.

Discussion Point: Redefining Musicianship?

• Do you agree that using tools like Suno, involving intervention and curation, constitutes a form of musicianship?
• How does this “AI-assisted musicianship” compare to traditional musicianship (instrumental skill, composition theory, performance practice)?
• Is “curation” a valid form of musical creativity in this context?

Conclusion: Democratization & Impact

• Suno exemplifies music production shifting from labor-intensive to technology-intensive.
• Furthers the democratization of music production access.
• Contributes to the democratization of musicianship itself (redefining who counts as a musician).

Conclusion: Future Implications

• Industry Impact: Plausible that AI music generators replace human composers/musicians for functional music needs (film, ads, games).
  • Suno’s Pro/Premier ownership rights support this.
• Raises ongoing questions about value, creativity, and the future role of human artists.

Final Discussion Point: Reflections & Futures

• What is your single biggest takeaway or most surprising thought from this discussion?
• How might tools like Suno change your personal relationship with creating or listening to music?
• Looking ahead, what ethical considerations or societal conversations about AI music seem most urgent?