Threads of extinction: How AI tapestries preserve the stories of vulnerable species

As machine learning models become more integrated into creative practices, their influence extends beyond digital canvases to traditional crafts such as textile weaving. The research paper "Impossible Evolutions: Textillic Thinking with Machine Learning Models" by Kate Geck, published in AI & Society, introduces the concept of "textillic thinking" as a framework for understanding the relationship between artificial intelligence (AI) and creative processes. This approach explores the intertwined nature of human and machine collaboration, using textile metaphors to highlight AI’s role in generating new artistic expressions.

Textillic thinking: A new perspective on AI creativity

Textillic thinking refers to the conceptual practice of viewing AI-generated outputs as woven artifacts, where machine learning models and human creators interlace their contributions into a cohesive whole. Geck’s research draws from historical and philosophical perspectives, linking textile production to computational systems. From the Jacquard loom - an early precursor to computer programming - to the algorithmic complexities of generative adversarial networks (GANs) and diffusion models, textiles have long provided a structural metaphor for digital logic and pattern recognition.

The study examines how AI can be both a creative tool and a collaborator in artistic practices. By training machine learning models on images of endangered Australian butterflies and wildflowers, the Impossible Evolutions project generated unique digital patterns that were later transformed into woven tapestries. These machine-imagined species highlight the potential of AI in speculative design, offering an artistic response to biodiversity loss while simultaneously demonstrating AI’s ability to produce novel yet contextually rich visual representations.

Machine learning and the materiality of art

One of the key arguments in the research is the importance of understanding AI’s materiality - that is, recognizing that machine-generated images and data-driven designs are not simply abstract outputs but tangible artifacts shaped by algorithmic and human influences. The study discusses "algomateriality," a term that encapsulates the interplay between algorithms and the material world. Machine learning models, much like fibers in a textile weave, interact dynamically with datasets, user inputs, and computational biases to form emergent patterns.

The Impossible Evolutions project emphasizes this interplay by demonstrating how machine learning can create digital designs that are then physically realized through textile weaving. The tapestries act as a bridge between the virtual and the material, offering a critique of how AI systems process, manipulate, and reproduce information. This methodology encourages artists and designers to consider not only the aesthetic potential of AI but also its ethical implications, particularly in terms of data sourcing and environmental sustainability.

Collaboration between human and machine intelligences

Rather than viewing AI as a replacement for human creativity, the research positions it as a collaborative entity that can expand artistic possibilities. Geck explores how artists working with AI are engaged in an iterative, reciprocal process akin to the interwoven threads of a textile. The research suggests that AI’s generative processes can be guided and curated by human intent, emphasizing the agency of the artist in shaping the final outcome. This collaboration raises critical questions about authorship, originality, and the role of machine learning in artistic practice.

Through the lens of Impossible Evolutions, the study presents three speculative narratives based on AI-generated ecological imagery. Each tapestry tells a story of an endangered species and its ecosystem, drawing attention to broader themes of conservation, technology, and symbiosis. By weaving these narratives into physical textiles, the project underscores how AI can function as both a creative partner and a storytelling medium, capable of producing meaningful and culturally resonant works.

Future directions: AI, textiles, and ethical considerations

The paper concludes by advocating for a more critical engagement with AI in creative domains, urging practitioners to consider the ethical dimensions of machine-generated art. Issues such as data transparency, intellectual property, and environmental impact remain central to discussions about AI’s role in artistic production. Geck proposes that textillic thinking offers a valuable framework for navigating these challenges, encouraging artists and technologists to approach AI with a mindset of care, collaboration, and material awareness.

Looking ahead, the integration of AI into textile arts and other traditional crafts presents exciting possibilities for innovation. Future research could explore how AI-driven weaving techniques might enhance sustainable textile production, how machine learning models can better capture cultural and historical textile patterns, and how AI can contribute to preserving endangered artistic practices. By continuing to weave together human ingenuity and machine intelligence, textillic thinking provides a rich and evolving paradigm for exploring the intersections of art, technology, and ethics.