Inverted Ekphrasis and Hallucinating Stochastic Parrots: Deleuzean Insights into AI and Art in Daily Life
DOI:
https://doi.org/10.54103/2039-9251/27840Parole chiave:
large language models, Gilles Deleuze, art, everyday lifeAbstract
This study explores the potential of contemporary large language models (LLMs) to achieve Gilles Deleuze’s goal of integrating art into daily life. Deleuze’s philosophy, with its focus on creative repetition, finds a parallel in LLMs, which replicate and innovate artistic styles by transforming text prompts into various artistic expressions. Although LLMs can very effectively blend and mimic these styles, they remain mere tools – as suggested by the metaphor of a stochastic parrot; the true creative force remains the human artist. Currently, these models are experimental, and their artistic outputs, while impressive, are still viewed as pastiches rather than genuine art. Nonetheless, we can expect advances that may soon enable artists to use LLMs to create genuine artworks.
Downloads
Riferimenti bibliografici
Bender, E. M., et al., On the Dangers of Stochastic Parrots. Can Language Models Be Too Big? in Association for Computing Machinery, FAccT ’21. Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, Association for Computing Machinery, New York 2021, pp. 610-623, https://doi.org/10.1145/3442188.3445922.
Benjamin, W., The Work of Art in the Age of Mechanical Reproduction (1935), in H. Arendt (ed.) Illuminations, translated by H. Zohn, Schocken Books, New York 1968, pp. 217-251.
Chelli, M., et al., Hallucination Rates and Reference Accuracy of ChatGPT and Bard for Systematic Reviews. Comparative Analysis, in “Journal of Medical Internet Research”, 26, 2024, https://www.jmir.org/2024/1/e53164.
Christiano, P., et al., Deep Reinforcement Learning from Human Preferences, in “arXiv”, 1706.03741, 2023, https://doi.org/10.48550/arXiv.1706.03741.
Deleuze, G., Difference and Repetition (1968), translated by P. Patton, Columbia University Press, New York 1994.
Deleuze, G., Proust and Sign (1964), translated by R. Howard, Athlone Press, London 2000.
Fišerová, M., Touching and Retouching. Question of Authenticity in Social Networking, in S. Wrobel and K. Skonieczny (eds) Living and Thinking in the Postdigital World, Universitas, Kraków 2021, pp. 111-126.
Hull, G., Transformer Models, Iterability, and Language (Part 1), in “New APPS: Art, Politics, Philosophy, Science”, 10 July 2024, https://www.newappsblog.com/2024/07/transformer-models-iterability-and-language-part-1.html.
Kozloff, M., Pop Art, Culture, and Rebellion, in “Artforum”, 1(3), 1962, pp. 27-33.
PromptDervish, PromptDervish Pastiches: Pop Art Reimagined with AI, in “Medium”, 6 January 2024, https://pub.aimind.so/promptdervish-pastiches-pop-art-reimagined-with-ai-050f81564a58.
Ray, P. P., ChatGPT. A Comprehensive Review on Background, Applications, Key Challenges, Bias, Ethics, Limitations and Future Scope, in “Internet of Things and Cyber-Physical Systems”, 3, 2023, pp. 121-154, https://doi.org/10.1016/j.iotcps.2023.04.003.
Yenduri, G., et al., Generative Pre-Trained Transformer. A Comprehensive Review on Enabling Technologies, Potential Applications, Emerging Challenges, and Future Directions, in “arXiv”, 2305.10435, 2023, https://doi.org/10.48550/arXiv.2305.10435.
Dowloads
Pubblicato
Come citare
Fascicolo
Sezione
Licenza
Copyright (c) 2025 Lenka Lee

Questo lavoro è fornito con la licenza Creative Commons Attribuzione - Condividi allo stesso modo 4.0.
Gli autori che pubblicano su questa rivista accettano le seguenti condizioni:
1. Gli autori mantengono i diritti sulla loro opera e cedono alla rivista il diritto di prima pubblicazione dell'opera, contemporaneamente licenziata sotto una Licenza Creative Commons - Attribuzione - Condividi allo stesso modo 4.0 internazionale che permette ad altri di condividere l'opera indicando la paternità intellettuale e la prima pubblicazione su questa rivista.
2. Gli autori possono aderire ad altri accordi di licenza non esclusiva per la distribuzione della versione dell'opera pubblicata (es. depositarla in un archivio istituzionale o pubblicarla in una monografia), a patto di indicare che la prima pubblicazione è avvenuta su questa rivista.
3. Gli autori possono diffondere la loro opera online (es. in repository istituzionali o nel loro sito web) prima e durante il processo di submission, poiché può portare a scambi produttivi e aumentare le citazioni dell'opera pubblicata (Vedi The Effect of Open Access).