AI and the Collective Subconscious: From Surrealist Automatism to Digital Dream Machines

Književstvene – Association of Citizens

Can a machine dream? And if it can, whose dreams is it dreaming?

In this Versopolis Traineeship Class, Diana Dukanac and Valentina Bošković Marković (from Književstvene – Association of Citizens) explore the fascinating territory between Surrealism, automatic writing and generative AI, asking what happens when century-old experiments with the subconscious encounter today's digital ‘dream machines.’

Across two interconnected lectures, the class begins with the Surrealists’ desire to escape rational control and allow language to emerge freely through automatic writing. Techniques such as uninterrupted writing, altered states of attention and spontaneous association challenged traditional ideas of artistic intention and authorship: if the writer gives up control, who – or what – is actually creating?

The same question acquires new urgency with generative AI. Drawing on ideas of the collective subconscious, literary influence, textual uncertainty and AI as a ‘dream machine,’ the class asks whether machine-generated language might represent a new kind of computational automatism – capable of reproducing dream logic, unexpected associations and surreal imagery without possessing dreams, memories or a subconscious of its own.

Theory is then put into practice through an experimental workshop. Participants create their own automatic writing, transform their texts into AI-generated images, and then ask AI to turn those images back into new texts – a contemporary digital echo of the Surrealist Exquisite Corpse. The resulting human and machine texts are compared through their imagery, symbolism, emotional intensity and associative logic.

The experiment reveals a compelling tension. Human automatic writing tends to be spontaneous, irregular and emotionally immediate, while AI can imitate dream-like language with striking fluency but often produces something more polished and controlled. The class therefore opens onto larger questions about creativity, authorship, originality and the cultural material from which AI learns.

Rather than asking simply whether AI can replace human creativity, this Traineeship Class invites us to ask something more interesting: What can our encounters with AI reveal about the strange, unpredictable processes through which humans themselves create?

And if AI really is a digital dream machine – whose dreams does it contain, and whose voices are missing?

Further Reading:

• André Breton, The Automatic Message (1933)
  If you’d like to explore the origins of automatic writing in more depth, start with Breton himself. In The Automatic Message, you can encounter the Surrealist fascination with automatism, trance, chance and the possibility of allowing language to emerge beyond conscious control.
• T. S. Eliot, ‘Tradition and the Individual Talent’ (1919)
  If you’re curious about the idea of the artist as a medium rather than an entirely autonomous creator, Eliot’s famous essay is a fascinating companion to this lecture. Eliot explores the relationship between individual creativity and the body of literature that comes before it, arguing that the poet’s work emerges through a complex interaction with tradition. His ideas about artistic “impersonality” and the surrender of the individual self offer an intriguing point of comparison with both Surrealist automatism and AI-generated texts.
• Harold Bloom, The Anxiety of Influence: A Theory of Poetry (1973)
  Bloom’s theory of poetic misprision looks at how writers transform, reinterpret and sometimes deliberately misunderstand their predecessors in order to create space for something new. You may wonder whether an AI’s strange transformations of existing language might be understood in a similar way.
• Kenneth Goldsmith, Uncreative Writing: Managing Language in the Digital Age (2011)
  If you’re interested in what happens to ideas of originality and authorship when we have access to enormous amounts of existing language, Goldsmith offers a provocative place to continue. His discussion of appropriation, copying, recombination and the creative rearrangement of existing texts provides an interesting bridge between avant-garde literary practices and contemporary AI-assisted writing.
• Arthur I. Miller, The Artist in the Machine: The World of AI-Powered Creativity (2019)
  If you’d like to move from literary theory into contemporary AI, Miller offers an accessible exploration of machine-generated art, literature and music. As you read, you can consider the question at the heart of our discussion: when a machine produces something that appears creative, where exactly does the creativity reside – in the machine, its training data, its programmer, its user, or some combination of all of these?
• Kate Crawford, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (2021)
  If the metaphor of AI as a ‘dream machine’ interests you but you also want to look behind the metaphor, Crawford's book is an important counterpoint. It invites you to consider the physical resources, data, labour and power structures that make apparently immaterial AI systems possible – and to ask whose voices, work and experiences become part of the machine’s ‘dreams.’
• Billy Perrigo, ‘Exclusive: OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic,’ TIME (2023)
  For a shorter and more accessible exploration of AI labour, this article offers a concrete case study of the human work involved in training and moderating AI systems. It provides a useful starting point for thinking about what can disappear from view when we imagine AI as an autonomous or almost magical form of intelligence.
• N. Katherine Hayles, Bacteria to AI: Human Futures with Our Nonhuman Symbionts (2025)
  If you want to take the question of artificial intelligence further into literary theory and philosophy, Hayles offers a fascinating contemporary perspective. Her work explores different forms of intelligence and asks us to rethink what we mean by cognition, agency and meaning when humans are no longer the only actors involved.
• Dav Pilkey, Dog Man series
  And finally, if you need a reminder that creativity does not always have to be serious, philosophical or theoretically defensible, or simply have children, spend some time with Dog Man. It may seem like the odd one out here, but Pilkey’s wonderfully chaotic combination of comics, storytelling, visual experimentation, absurdity and play is a joyful reminder that creativity can emerge through unexpected combinations, mistakes, transformations and sheer fun. And honestly, sometimes a heroic half-dog, half-human superhero is exactly what you need after Derrida.