At DePaul University’s recent AI in the Arts Symposium, a diverse panel of artists and technologists gathered to tackle one of the most pressing questions in contemporary art: How can AI serve as a creative collaborator rather than a replacement for human artists? The panel, moderated by Professor Rob Steel from DePaul’s School of Cinematic Arts, brought together voices from Northwestern University, the University of Chicago, the Art Institute of Chicago, and DePaul to explore this evolving relationship between artificial intelligence and artistic practice.
The “Yes, And” Philosophy of AI in the Arts
Professor Bryan Pardo from Northwestern University’s Interactive Audio Lab opened the discussion with a powerful metaphor that would resonate throughout the panel. Holding up a clarinet, he reminded the audience that when recording technology emerged, the clarinet didn’t disappear—it got recorded. When sampling arrived, it got sampled. Now, with generative AI, the clarinet remains. His message was clear: AI should represent a “yes, and” approach to artistic creation, not an either-or proposition.
This philosophy manifested in Pardo’s demonstration of cutting-edge sound generation tools his lab has developed in collaboration with Adobe. Rather than simple text prompting, these tools allow sound designers to control AI generation through physical gestures and rhythm inputs. The demonstration showed how an artist could guide AI to generate a lion’s roar synchronized with video, maintaining creative control throughout the process. As Pardo explained, “Sound design isn’t going to go away. But if we design the tools right, the tools will not be to supplant us. They will be to support us.”
Pushing Beyond the Generic: Artists Taking Control
Jason Salavon, an Associate Professor at the University of Chicago who has been creating computational art for over three decades, brought a fine artist’s perspective to the conversation. His work demonstrates how artists can push AI systems beyond their comfort zones to create genuinely unique outputs. Through his studio’s experiments with what he calls “putting the model on acid”—injecting transformations that push diffusion models outside their normal rendering space—Salavon creates images that escape the “bland” quality often associated with AI-generated art.
His approach emphasizes customization over acceptance of default outputs. “I think of them as image hoses,” Salavon explained, describing how image generation models can become brushes for spraying narrative across larger compositional structures. This perspective transforms AI from a push-button solution into a sophisticated tool requiring artistic vision and technical expertise to yield meaningful results.
The Challenge of Maintaining Artistic Voice
Rufino Jimenez, an Art Specialist at the Art Institute of Chicago and Adjunct Professor at DePaul, offered a compelling demonstration of both AI’s potential and its limitations. Through a creative experiment where he asked ChatGPT to create a cinematic presentation about his journey into AI, he showed how generative models progressively lose coherence and detail with each iteration. What began as cinematic imagery devolved into generic, sometimes absurd outputs—a vivid illustration of why human guidance remains essential.
Yet Jimenez also shared how AI helped him complete a long-standing artistic project: creating a mathematical formula representing “all drawings that were ever made, and all drawings that will be made till the end of time.” AI’s ability to process complex mathematical concepts and generate proofs enabled him to realize an artistic vision that had been technically out of reach. This duality—AI as both limitation and liberator—captures the nuanced reality artists face when integrating these tools into their practice.
Learning from History: Photography’s Lessons for AI
Jessica Larva, Associate Professor at DePaul’s Art School, provided crucial historical context by drawing parallels between current AI anxieties and past technological disruptions in art. When photography emerged, critics dismissed it because “the camera does all the work.” Similar criticisms were leveled at digital art, with one publication telling computer art pioneer Charles Csuri they couldn’t imagine ever covering “electronics or computers in art.”
Yet photography didn’t end painting—it liberated it from the “tyranny of realism” and opened new creative horizons. Digital art similarly expanded our relationship to media and space. Larva argues that AI will follow a similar trajectory: “Artists will reject the parts of AI that they don’t like or don’t compete with, and they will find new ways to connect with other humans. They’ll differentiate. They’ll find new lanes and expand new horizons.”
The Ethics of Attribution and Influence
A recurring theme throughout the panel was the need for transparency in AI systems. Pardo’s lab has developed models that can identify which training data most influenced a generated output, returning what he calls “ethical agency” to users. This approach acknowledges that artistic creation has always involved influence and borrowing—from Beethoven studying Mozart to contemporary sampling artists—but insists on making these influences visible and attributable.
The panelists agreed that the current opacity of AI systems, driven partly by legal concerns around copyright, prevents artists from making informed ethical choices about their work. By exposing the genealogy of AI-generated content, artists can move from being “ignorant appropriators to informed borrowers,” understanding and acknowledging their creative lineage.
Tools for Artists, By Artists
Perhaps the most hopeful message from the panel was the emergence of AI tools specifically designed for and by artists. Rather than accepting commercial tools designed for mass appeal, artists and researchers are creating systems that preserve creative control, maintain individual voice, and support artistic workflow. From Pardo’s gesture-controlled sound generation to Salavon’s image transformation systems, these tools demonstrate that AI’s future in the arts need not be dictated by venture capital priorities.
The panel also emphasized accessibility. As AI lowers barriers to coding and technical implementation—what some call “vibe coding”—more artists can customize and create their own tools. This democratization of technical capability promises to put more power in artists’ hands, enabling them to shape AI to their creative vision rather than adapting their vision to AI’s limitations.
Looking Forward: The Avant-Garde Opportunity
The conversation concluded with a sense of cautious optimism. While acknowledging real concerns about attribution, authorship, and the flood of generic AI content, the panelists saw opportunity in the current moment. As new models emerge and artists experiment with novel applications, we’re witnessing what might be rapid-fire art movements—three-day cultural conversations sparked by new capabilities and creative responses.
For students and emerging artists, the message was clear: those who learn to use AI tools idiosyncratically, who push against their intended purposes, who layer physical and digital creation, and who maintain their unique artistic voice while leveraging AI’s capabilities will define the next era of creative practice. The challenge isn’t whether to use AI, but how to use it in ways that amplify rather than diminish human creativity.
As the symposium demonstrated, the conversation about AI in the arts is far from settled. But by bringing together technologists who think like artists and artists who embrace technology, DePaul’s AI Institute is fostering the kind of interdisciplinary dialogue necessary to ensure that AI serves as a true creative collaborator—one that expands artistic possibilities while preserving the human elements that make art meaningful.