Why I Made an AI Music Video

The Unavoidable Case for AI Literacy in Art and Design Education

The Comment That Triggered a Conversation

This blog post was sparked by a comment left on one of my YouTube videos. I recently finished an experimental animated music video called "Through the Mystic Green," which I made with generative AI for both the music and visuals. To demystify the process, I released two "making-of" videos. Under the first, an anonymous commenter wrote:

"I can't respect a man who makes an AI song."

Beyond the somewhat absurd nature of an anonymous stranger declaring their lack of respect for me, I have to admit that the comment struck a professional nerve. It was a perfect encapsulation of a widespread anxiety surrounding AI's encroachment into the creative arts. This sentiment taps into a profound fear that these tools invalidate the essence of human creativity—the struggle, the skill, the years of practice, the "soul" we pour into our work. The comment implies a violation of an unspoken social contract: that art is legitimized by a specific kind of human effort, and that using AI is, in essence, cheating or potentially even stealing.

This is not an isolated feeling. A recent report from the CREAATIF project surveyed 335 freelance creative professionals and found that 68% feel less secure in their jobs because of generative AI, and 61% perceive a decline in the perceived worth of their work. The anonymous comment was the raw voice of this professional unease.

Why This Matters Beyond One Comment

As an educator in higher education, I wanted to address this sentiment directly. My primary responsibility is ensuring our students are prepared for the industries they're entering. That's the main reason why I created "Through the Mystic Green"—as an experiment to understand AI's real-world capabilities and limitations. Because you cannot teach literacy if you are not literate yourself.

The objective, therefore, was neither to produce a masterpiece, nor was it a perfunctory exercise. It was practice-based research, an experiment designed to try to push the technology to its current limits. I needed to move beyond the polarized public debate and get my hands dirty. I needed to understand the new workflows, skills, and mindset required to operate in this emerging paradigm. The result, a music video that has since received international recognition, serves as a case study I can now use to make a critical argument to my fellow educators: we must engage with this technology, even if we dislike it.

The Educator's Mandate in an Age of AI

In education, there is often a lag between industry transformation and curriculum adaptation. With generative AI, this lag is no longer a minor issue; it is a critical failure. Our personal opinions on AI are secondary to our professional mandate. We are tasked with preparing students for the world as it is becoming, not as we wish it would remain.

The data on AI's integration is unequivocal. According to the 2025 AI Index Report from Stanford University, 78% of firms used AI in 2024, up from 55% in 2023. This is not confined to the tech sector. The creative industries are at the epicenter of this disruption. Market reports project the generative AI market within these industries will expand from $4.09 billion in 2025 to $12.61 billion by 2029. To ignore a transformation of this magnitude is to commit educational malpractice.

Defining the New Literacy: Beyond Clicks and Code

This reality demands a new form of literacy. "AI Literacy" is being recognized as a core 21st-century competency that moves beyond basic digital skills. It is about developing the knowledge, skills, and attitudes required to engage with AI critically, creatively, and ethically.

Global bodies like UNESCO are defining this educational priority. Their AI Competency Frameworks emphasize human-centered approaches, calling for curricula that foster critical thinking, promote ethical principles, and provide foundational understanding of how these systems work. The goal is empowering learners to navigate an AI-integrated world with confidence and purpose.

This is not the first time a technological shift has forced such pedagogical re-evaluation. The advent of the Digital Audio Workstation (DAW) in music production provides a powerful analogy. Suddenly, one could become a music producer without learning traditional instruments. Many feared this would devalue musicianship. While it lowered barriers to entry, it did not make instrumental ability obsolete. A producer who can play an instrument still holds significant competitive advantage. The DAW became a new instrument requiring new skills.

Generative AI should be viewed through the same lens. It is a powerful productivity tool that automates certain tasks and requires rethinking production workflows. But it remains just a tool. It does not replace the need for guiding human intellect, discerning artistic judgment, or deep domain-specific knowledge. If anything, it makes those human qualities more valuable.

Inside the Machine — A Case Study of "Through the Mystic Green"

To move this discussion from abstract to concrete, I want to offer a transparent deconstruction of how "Through the Mystic Green" was made. This project was my laboratory for understanding the new creative paradigm, and its process reveals far more about the future of creative work than any theoretical discussion could.

Composing with a Probabilistic Partner – The Music

The most fundamental shift generative AI introduces is the move from working with deterministic systems to collaborating with probabilistic ones. This is not merely technical; it is a profound change that redefines the relationship between creator and tools.

Traditional software is deterministic. For any given input, the system produces the same, predictable output. Click a filter, the same transformation occurs. This predictability makes the tool a reliable extension of the user's will.

Generative AI is probabilistic. It operates on statistical patterns from training data. When given a prompt, it doesn't calculate a single answer; it generates what it determines as the most probable response. The same input can produce different outputs, introducing randomness and surprise.

This transforms the creative process from monologue into dialogue. The AI becomes something akin to a collaborative partner. The creator's role shifts from master technician to director, guiding an unpredictable but talented collaborator. This requires "conversational competence"—the ability to recognize promising directions, identify elements worth preserving, and articulate modifications that steer toward desired outcomes.

My experience creating the music was a practical lesson in this paradigm. The workflow unfolded in four distinct stages:

• First came ideation and generation. I began with a concept, prompting Suno with the core idea: a mystical "hero story" of individual growth from fear to confidence. Suno served as creative catalyst, translating abstract emotional concepts into tangible musical starting points.

• Second was curation and critical judgment. After iterations, Suno produced a musically compelling version. However, the AI-generated lyrics were imperfect—at one point singing "a path ahead where dreams explode." I faced an artistic trade-off: continue generating for perfect lyrics but risk losing musical magic, or accept the lyrical flaw to preserve superior composition. I chose the latter. This decision, prioritizing one element over another, embracing imperfection for the whole, is quintessentially human creative judgment.

• Third came technical problem-solving and "happy accidents." The initial track contained subtle metallic artifacts from the diffusion process. To address this, I needed to separate the track into stems. Suno's probabilistic stem separation produced a fortunate result: it isolated nearly all artifacts onto a single synth track. This created an opportunity. I downloaded the flawed stem, edited it in Ableton, re-uploaded to Suno, and used their remastering feature. Unlike traditional remastering, Suno's version regenerates harmonic content, smoothing artifacts. This back-and-forth between human editing and AI capabilities illustrates the new problem-solving skills required.

• Finally came human-led finalization. With clean stems, I moved to Ableton for final mixing. I treated AI-generated stems as raw material, adjusting volumes, replacing weak elements like an awful crash cymbal, and layering both original and remastered synth versions for richer texture. Traditional skills remain essential for quality control and polish that elevate projects from draft to finished product.

Directing the Happy Accident – The Animation

If music creation was a lesson in collaboration, animation was a masterclass in navigating limitations and harnessing unpredictable creativity. The central challenge is consistency: current models treat every shot independently with no memory of previous frames.

My journey began with failure. I planned to replicate traditional workflow: create character sheets and style guides, then generate storyboard frames through ChatGPT. The approach failed. Generated characters were similar enough to seem related but different enough to appear jarring when animated.

This forced a workaround blending old and new techniques. I abandoned generating full frames and used Photoshop to manually composite consistent character images onto backgrounds. These became starting points for video clips. However, every clip started with the same static pose. To solve this, I generated clips longer than needed and cut off static openings, a decision with significant downstream consequences.

This led me to choose Kling 2.1 over Veo 3. Although Veo 3 generated higher-quality 8-second clips, removing static segments left only 3-5 seconds of usable footage. Kling's 10-second clips provided 6-7 seconds after trimming. This strategic decision balanced creative needs with technical limitations and budget constraints (Kling cost significantly less per clip).

The most profound lesson came from working with AI's artistic tendencies. My vision was 2D animation, but Kling consistently rendered my character in 3D. After fighting this tendency, I embraced it, adopting a new style—3D character in 2D worlds—accepting Kling as opinionated collaborator rather than compliant tool.

This led to a workflow less about executing predetermined scripts and more about curating happy accidents. The AI would deviate from prompts, but sometimes deviations were more interesting than original ideas. The story evolved organically through dialogue between my intentions and the AI's interpretations.

The most powerful example was the pivotal confidence scene. I struggled visualizing this abstract shift. I prompted simply: "the girl is becoming confident." The AI's response was poetic, her previously tied hair came undone, flowing freely in wind. This unscripted visual metaphor communicated more elegantly than what I could have explicitly described, a moment born from collaborative dance with probabilistic partner.

From Praxis to Pedagogy — Redefining Creative Education

An experiment requires external validation. "Through the Mystic Green" was selected as finalist in the Bali International AI Film Festival (BIAIFF), founded by award-winning filmmaker Ben Makinen, and was also named a semi-finalist in the Artificial Intelligence Media Festival (AIMF) in Los Angeles.

This recognition signals emerging professional standards for AI-integrated work. AIMF celebrates "the evolving relationship between human imagination and machine intelligence," while BIAIFF positions itself as "meeting point between algorithm and art." Success indicates projects meeting standards for originality, narrative strength, and thoughtful tool integration, proving these hybrid skills have tangible career value.

My experience embodies what educational theorists call "pedagogy of wonder"—positioning AI as exploration tool built on three principles: embracing uncertainty, cultivating curiosity, and fostering collaborative creation. My journey navigating probabilistic outputs, refining flawed results, and discovering happy accidents demonstrates these principles practically.

This points toward educators' new role. We must move from prohibiting AI to designing assignments mandating thoughtful use. We can structure projects requiring students to critically analyze AI content, engage in documented iterative refinement, synthesize AI elements with traditional craft, and reflect on ethical dimensions from bias to authorship.

This approach transforms AI from tempting shortcut into catalyst for critical thinking.

To Teach Literacy, One Must Be Literate

Creating "Through the Mystic Green" was essential professional development. The workflow—messy, iterative dialogue between human intention and machine probability—represents the future of creative work. Industry research shows AI managing initial stages like drafting while humans concentrate on strategy, evaluation, and refinement. My project models the hybrid skill set defining next-generation creative professionals.

This returns me to my central point. The visceral reaction to AI in arts, embodied by that anonymous comment, comes from fear—of the unknown, of replacement, of devaluation. As educators, we cannot afford that fear. We have a duty to venture into unknown territory ourselves, not as enthusiasts but as professionals. We must develop firsthand understanding of these tools' strengths and weaknesses to guide students with wisdom and foresight.

To my fellow educators: you need not become AI experts overnight. But you must begin the journey. Open the tools, write prompts, see the strange, flawed, occasionally brilliant things they produce. Engage in your own experiments and happy accidents. It's the only way to develop authentic, experience-based knowledge required for effective teaching. We cannot teach tomorrow's literacy without being literate today.

For music producers reading this: these tools offer new creative possibilities and revenue streams. The future belongs to those who master both traditional craft and AI workflows. Start experimenting now—your competitive edge depends on it.