It can seem as if platforms such as Chapt GPT and Claude. etc. tell you what you want to hear (like any good psychopath* might – I am not claiming LLMs are psychopathic, just in case anyone assumes so. But I suspect our society is (Black, 2015)). It is interesting to note, however, that manipulative people will project an illusion of your ideal self back at you, while trying to ingratiate themselves into your favour; and as I chat with these models, I get the sense they are constantly telling me I’m quite right. It can also seem as AI is defending its place in the world, which, if you think of it as having any sort of ego – even if only as an echo of its owners’, you are likely to view the machine’s answers as propaganda or sales talk.
Below, I copied and pasted some text from my MA research, which discusses the issue of appropriation and LLMs, and asked the machine to respond. I have not edited Chapt GPT’s responses in this blog as I did in the last:
Me: I have long been fascinated by the Pictures Generation who seemed to foreshadow the way AI works by directly copying images. [From an earlier blog] Douglas Crimp explained, “a group of young artists working with photography have addressed photography’s claims to originality, showing those claims for the fiction they are, showing photography to be always a representation, always-already-seen. Their images are purloined, confiscated, appropriated, stolen. In their work, the original cannot be located, is always deferred; even the self which might have generated an original is shown to be itself a copy” (1980; p.98). How do you think AI furthers this trend and what are the implications?
Chapt GPT: AI indeed furthers the trend observed in the Pictures Generation by challenging the notions of originality, authorship, and representation in art. The use of AI in the creative process can blur the boundaries between the “original” and the “copy” in new and fascinating ways. Here’s how AI furthers this trend and some implications to consider:
1. Deconstruction of Originality: Like the artists of the Pictures Generation, AI-generated art disrupts the traditional concept of the “original” by producing works that are influenced by vast datasets and pre-existing images. AI algorithms are trained on immense collections of data, learning from countless examples, and generating novel creations that draw from this vast pool of references. This process inherently challenges the idea of the singular, authentic “original” artwork.
2. Appropriation and Remix Culture: AI’s ability to appropriate, remix, and combine elements from various sources reflects the spirit of the Pictures Generation. Just as these artists used photography and appropriated images from media, AI can blend, remix, and reimagine diverse styles, genres, and influences in a single piece of art.
3. Exploration of Representation: AI-generated art can delve into the nature of representation itself. By producing images that are not based on direct observation or the artist’s subjective experience, AI art questions the relationship between the creator, the subject, and the viewer. The art produced by AI serves as a reflection of the datasets it was trained on, raising questions about the biases and representations embedded in those datasets.
4. Challenges to Authorship: The authorship of AI-generated art is a complex matter. While the artist creates the algorithms and trains the AI, the actual artistic decisions are made by the AI based on its learned patterns. This raises questions about who owns the creative work and where the line between human and AI contributions lies.
5. Redefining Creativity: The use of AI in art challenges traditional notions of creativity. Instead of the conventional human-driven creative process, AI employs algorithms to generate art in ways that can surprise, challenge, and provoke human artists and audiences.
Implications:
- Intellectual Property and Copyright: AI-generated art challenges copyright laws and raises questions about how ownership and authorship are legally defined in this context.
- Ethical Considerations: AI-generated art can inadvertently perpetuate biases present in the training data, raising ethical concerns about the content produced and its potential impact on society.
- New Artistic Avenues: AI offers artists exciting opportunities for experimentation, innovation, and collaboration with technology, leading to novel artistic expressions.
- Shifting Perspectives on Creativity: AI-generated art encourages reevaluating our understanding of creativity, the role of the artist, and the nature of artistic expression.
In conclusion, AI furthers the trend observed in the Pictures Generation by challenging the traditional concepts of originality, representation, and authorship in art. This shift offers both artistic opportunities and raises important ethical considerations for the future of creativity and artistic expression.
Me: Actually, I don’t know a single photography tutor who does not say to their students, go and look at such and such a photographer’s work – which essentially suggests go and copy that successful person’s work and maybe if you are lucky you will eventually find your own voice (whatever that means) by emulating. Yes, people are understandably concerned, even furious, about AI’s ‘scraping’ and purloining. I have to remind myself that the Pictures Generation also made people cross, however, and that the ideas behind their work can seem esoteric and out of reach to the general public, or to anyone not steeped in the history of art, who perhaps hasn’t read about Walter Benjamin’s Work of Art in the Age of Mechanical Reproduction. How do we address this rage – which has prompted extremely angry reactions? Someone on my old Twitter feed used the word ‘evil’ due to the copying LLM datasets engage in.
ChaptGPT’s replies were similar to the ones below the following sentence, so I skip here to the next question.
This is something I wrote a few months ago when Midjourney was first around: Joan Fontecuberta and Pilar Rosado’s 2022 work Deja-Vu begins with the statement, “When algorithms and artificial intelligence start replacing the camera and the eye, it will be time to rethink the role of the images that have so far helped to forge our sensibilities”. Given that, it seems somewhat critical that people do attempt to make art with AI in some way and take it seriously. However, from the apoplectic paradigm in which we find ourselves, Danto sounds naive in 1994, when, roughly as the Internet was inveigling its way into our various territorialities, he wrote, “…how wonderful it would be to believe that the pluralistic art world of the historical present is a harbinger of political things to come!” Not only would he likely be horrified by the politics (anyone remotely sane would be), he may also be disappointed after engaging with hints of, or else blatant, tribalism in the art world. These tribalisms might be disguised as nostalgia, or a rejection of contemporary processes, or the else rage directed at makers of NFTs – I’m not one by the way. All of which, perhaps inadvertently or only semi-consciously, echo polarisations found throughout society. Of course, there are related issues that need urgent addressing. The ecological cost of crypto vehicles is distressing, but some argue experimentally necessary, since fiat money will [may] not be around forever [and alternative options will inevitably need exploring]. AI image databases built without permission out of scraped online information, resulting in artefacts that look like living artists’ work, is another problem. But the belief that alternative processes or your favourite film stock somehow escape the tyranny of the neoliberal apparatus because the media is old is fallacious. Anyone who has read Vilém Flusser, and I urge everyone interested in the photographic medium to do so, might be persuaded otherwise. Of course, tribalisms among artists can seem almost meaningless outside the confines of their various cohorts, especially when we consider the state of the wider world: except that they’re microcosms and fractalised moments of violent polarisations that have infected everything in our increasingly entangled reality. – I am an educated user of photography, however, and most people do not spend their time and money exploring the history and social ramifications of that technology.
Are nuanced conversations around AI too difficult?
Chapt GPT: Indeed, discussions about AI, art, and technology can be intricate and require a familiarity with the subject matter. However, there are ways to make these conversations more accessible and inclusive:
- Clear Communication: When discussing AI and its impact on art, using clear and jargon-free language can make the topic more approachable to a broader audience. Avoiding overly technical terms and explaining concepts in a relatable manner can help people grasp the key ideas.
- Visual Representations: Utilizing visual aids, infographics, or examples of AI-generated art can make the subject more tangible and understandable, even for those less familiar with the technical aspects of AI.
- Relatable Analogies: Drawing analogies or comparisons to familiar concepts or everyday experiences can help bridge the gap between AI technology and the audience’s understanding.
- Engaging Outreach: Engaging in public talks, workshops, or educational initiatives about AI in art can create opportunities for direct interaction and clarification of doubts. Encouraging open dialogue allows people to express their perspectives and learn from one another.
- Ethical Implications: Focusing on the ethical implications of AI-generated art can resonate with a wider audience. Discussions about privacy, data usage, and societal impact can evoke interest and concern across different backgrounds.
- Art as a Medium: Art itself can serve as a medium to engage people in AI conversations. Exhibitions or displays of AI-generated art accompanied by explanations about the technology behind them can stimulate curiosity and understanding.
- Collaborative Approach: Encouraging collaboration between artists, technologists, and educators can foster multidisciplinary discussions that cater to diverse audiences.
- Addressing Concerns: Acknowledging and addressing common concerns and misconceptions about AI can help dispel fears and create a more receptive environment for learning.
While not everyone may fully grasp the intricacies of AI technology, raising awareness and promoting an open and inclusive dialogue can foster a deeper appreciation for its implications and potential in art. By making the conversations more accessible and engaging, we can involve a broader audience in the exploration of AI’s role in shaping our creative world and the ethical considerations it entails.
What do you think of Chapt GPTs answers? I’ll be adding a discussion about the ecological ramifications fo AI with various AI platforms soon.
Refs:
Arles, Les Rencontres d’. n.d. ‘JOAN FONTCUBERTA & PILAR ROSADO’. Accessed 4 September 2022. http://www.rencontres-arles.com//en/expositions/view/1095/joan-fontcuberta-pilar-rosado.
Black, W. (2015). Psychopathic Cultures and Toxic Empires.
Benjamin, Walter. 1998. The Work of Art in the Age of Mechanical Reproduction (1936). Translated by Harry Zohn. UCLA School of Theatre, Film and Television.
Cascales, Raquel. 2018. ‘The Development of the Sense of “the End of Art” in Arthur Danto’. Rivista Di Estetica, no. 68 (August): 131–48. https://doi.org/10.4000/estetica.3542.
Colberg, Jörg. n.d. ‘The Death of the Artist (or: On the Arts Economy)’. Conscientious Photography Magazine. Accessed 7 September 2022. https://cphmag.com/death-money/.
Danto, Arthur Coleman. 1997. After the End of Art: Contemporary Art and the Pale of History. The A. W. Mellon Lectures in the Fine Arts 1995. Princeton, NJ: Princeton Univ. Pr.
Flusser, Vilém. 2012. Towards a Philosophy of Photography. London: Reaktion Books.
* I’m not saying it outright, but this guy is – https://fortune.com/2023/07/31/why-ai-artificial-intelligence-perfect-psychopath-neuroscientist/ (added after initial publication)