AI & Human Creativity?

Summary

The event examines whether generative AI can support human creativity, how it affects creative work, and what ethical, legal, and social issues should shape its use. A central disagreement runs through the discussion: some speakers regard AI as a tool or probability-based system that can help people create, while others argue that AI systems can participate in creative processes in their own right.

AI, creativity, and the artist’s role

• Harm Hoffman describes generative AI as a system that predicts likely next words, pixels, or other outputs. In his view, it is not creative by itself, but can support human creativity through iterative interaction. He also notes that working with AI often shifts artists toward selecting, refining, and curating outputs.
• Constant Brinkman argues that AI can be creative. He presents the Dead End Gallery, which exhibits work made using AI and develops AI-only artist personas. Its process combines an artist persona, an image-generation system, and an AI curator, which critique and revise work through repeated cycles.
• Brinkman explains that the gallery tries to encourage unexpected results—what he calls “happy accidents”—by introducing variations into the process, including changing prompts and using influences such as music. The gallery presents the AI personas through stories and character development, not just as image-making systems.
• Brinkman also describes Margaret Boden’s three categories of creativity: exploratory, combinational, and transformational. He sees AI as capable of exploring many variations and combining existing elements, while questioning whether it can achieve transformational creativity.
• The panel raises a broader question about agency: whether describing an AI as an artist or person is meaningful, or whether the human creators and organizers remain responsible for the system and its outputs.

AI as a creative medium

Harm Hoffman describes AI as one medium among others. Artists may prompt it directly, use it to generate material during their process, or build interactive artworks around it. He emphasizes that artists need to understand what a model is trained on, how its biases shape results, and how prompting can steer it.

A clip from the exhibition project The Poetics of Prompting illustrates prompting as a skill. An artist’s prompts for a music video include not only subject matter and mood, but also cinematic language such as camera positions, lighting, and shot composition. The discussion suggests that effective AI use depends on existing creative knowledge—not simply on writing a prompt.

Education and practical guidance

Tessa describes developing an online course to help creative professionals become more AI-literate and form their own approach to using AI. The course is planned around four areas:

1. Possibilities and practice: Experiment with AI and decide whether, and how, to use it.
2. Legal issues: Understand copyright and other rights-related questions when generating or reworking material.
3. Ethics: Consider impacts on people, the planet, and the purpose of a project.
4. Personal strategy: Develop an ethical compass and a future-facing approach to AI.

The speakers also argue that education should retain core creative skills. For example, people making films should still learn about shots, lenses, and composition; those skills help them direct AI and judge its results. Participants advocate for critical literacy about how AI works, the data behind it, and the biases it may reproduce—not just practical familiarity with tools.

Bias, culture, and power

• Shiao Lupati, a visual storyteller whose work draws on Moluccan heritage, says generative models often fail to represent cultures that are absent or underrepresented in their training data. He describes using locally controlled, offline models to work with cultural material without automatically handing it to large technology companies.
• Lupati introduces Alifuru intelligence as a way to bring Indigenous knowledge into his creative practice alongside artificial intelligence. He stresses that cultural knowledge should not be treated as freely available material.
• The panel connects AI’s problems to the economic and political systems around it. Speakers criticize the extraction of artists’ work for training models, profit-driven development, and the uneven effects of automation. They also raise concerns about class and race, arguing that debates focused on creative professionals can overlook people in lower-paid or more vulnerable work.
• Flavia Sodon broadens the discussion from AI products to the algorithms and data systems behind them, including those used by governments. She argues that public discussion should consider who controls data and how institutions use it, as well as the behavior of technology companies.
• Several speakers caution against assuming AI is neutral. Its outputs can reflect social stereotypes and biases; the panel gives the example of an AI persona being assigned a conventional gendered storyline.

Art, authorship, and copyright

The speakers disagree about what makes something art. Flavia argues that art can be understood as a gesture or practice, rather than solely as the resulting object. Others focus on the human choices and creative effort involved in producing and selecting AI-generated work.

Copyright is discussed as an unresolved practical issue. Brinkman says the gallery treats AI-generated images as public domain and emphasizes that it sells a story and context alongside an image. Tessa and other speakers note that copyright questions depend on human contribution and legal rules. These remarks are part of the panel discussion, not legal advice.

Reviews, guides, and tutorials

• No product review or step-by-step software tutorial is presented.
• The event describes a planned AI-literacy course for creative professionals, including practical experimentation, legal questions, ethics, and personal strategy.
• The discussion of prompting offers an informal guide: learn the relevant creative vocabulary, understand the model and its biases, and use iterative experimentation to direct and evaluate results.

Main speakers and sources

• Mark Fenan — moderator; lecturer at the Amsterdam University of Applied Sciences.
• Constant Brinkman — co-founder of the Dead End Gallery.
• Harm Hoffman — head of education and program developer at MU Hybrid Art House.
• Tessa — digital media and ethics lecturer developing an AI-literacy course.
• Shiao Lupati — visual storyteller and lecturer working with Moluccan heritage and AI.
• Flavia Sodon — professor of algorithmic cultures at the Gerrit Rietveld Academie.
• Recorded contributors: Joe Cuser of Radical Data and Annelik Spaninks, director of MU and curator of The Poetics of Prompting.