Research paper

I have been giving some thought to possible ideas for the research paper, and I want to keep it closely aligned with my study statement. When I was an undergraduate, I wrote my final-year dissertation on a topic that related directly to my studio practice, and I found that very helpful in shaping the contextual framing of the work. Here are some initial ideas:

1. Collage and AI generated images; historical precedent or false equivalence?  A theoretical comparison between collage and generative AI as modes of image construction through appropriation, recombination, and fragmentation. The focus would be on where the analogy might be useful and where it breaks down.
2. The question of origin in generative AI images. An examination of the ways in which AI unsettles ideas of source, authorship, and originality. This could look at training data, prompting, versioning, and model outputs as producing an image whose origin is distributed rather than singular.
3. Philosophical ontology versus technical ontology in AI
   A study of the tension between ontology as a philosophical inquiry into being and ontology as a technical structure of classification and relations in AI systems.
4. What kind of thing is an AI image? What is the status of the AI image? It is neither straightforward representation nor direct trace. Is it best understood as a composite, a simulation, a prediction, or something else?
5. Authorship after generative AI. What has generative AI done to ideas about the nature of authorship? This could examine the roles of dataset builders, platforms, model designers, prompt writers, and end users.
6. Visibility v invisibility in image construction. A piece on how different image systems reveal or conceal their own making. This could compare visible montage, photographic indexicality, and the apparent seamlessness of AI-generated imagery.
7. Extraction as an ontological condition of the AI image. Can the being of an AI image be understood without considering the conditions of its production? Are scraped datasets, hidden labour, water/ energy use, and corporate infrastructure constitutive rather than secondary considerations?
8. Iteration, repetition, and instability in machine-generated images. An examination of iteration and versioning in AI image making. What might Derrida’s ideas about trace and unstable origin tell us about the ways in which repeated generation alters the relation between image, source, and meaning?
9. Opacity and legibility in generative systems. A study of how AI systems (are designed to) resist interpretation. Looking at black-boxed processes, hidden datasets, and inaccessible decision structures. Asking what it means to theorise images whose conditions of emergence are partially concealed.
10. Is the term ‘AI slop’ a quality judgement or a symptom of ontological anxiety? Does it help clarify what is distinctive about generative AI imagery, or does it flatten important differences between kinds of images, uses, and contexts by turning them all into an undifferentiated category of visual waste? The piece would investigate both what the term reveals and what it obscures.