[Paper Review] AI Art is Theft: Labour, Extraction, and Exploitation, Or, On the Dangers of Stochastic Pollocks

Trystan S. Goetze|arXiv (Cornell University)|Jan 10, 2024

Cinema and Media StudiesEconomics, Econometrics and Finance46 references3 citations

TL;DR

This paper argues that generative AI art systems like DALL-E and Stable Diffusion constitute unethical labor theft by training on vast datasets of unconsented, uncredited artistic works, extracting value from human creators without compensation. It critiques the AI industry's labor extraction model as exploitative, drawing parallels to 'stochastic Pollocks'—random, algorithmically generated outputs that mimic artistic creation while erasing the real labor behind them.

ABSTRACT

Since the launch of applications such as DALL-E, Midjourney, and Stable Diffusion, generative artificial intelligence has been controversial as a tool for creating artwork. While some have presented longtermist worries about these technologies as harbingers of fully automated futures to come, more pressing is the impact of generative AI on creative labour in the present. Already, business leaders have begun replacing human artistic labour with AI-generated images. In response, the artistic community has launched a protest movement, which argues that AI image generation is a kind of theft. This paper analyzes, substantiates, and critiques these arguments, concluding that AI image generators involve an unethical kind of labour theft. If correct, many other AI applications also rely upon theft.

Motivation & Objective

• To examine the ethical implications of training generative AI models on vast datasets of copyrighted artistic works without consent or compensation.
• To investigate how the current AI art ecosystem enables systemic labor extraction from human artists, particularly in creative industries.
• To evaluate the artistic community's claim that AI-generated art constitutes 'theft' by analyzing the mechanisms of data harvesting and model training.
• To critique the narrative of AI as a neutral or beneficial tool, highlighting its role in displacing human labor and devaluing artistic labor.
• To argue that the ethical failures in AI art extend to other AI applications, suggesting a broader pattern of exploitation in AI development.

Proposed method

• Analyzes training data sources of major generative AI models (e.g., DALL-E, Midjourney, Stable Diffusion) to identify the unconsented inclusion of copyrighted artworks.
• Applies critical theory and labor ethics to frame AI training as a form of extractive labor exploitation, drawing on Marxist and feminist labor critiques.
• Compares AI-generated outputs to human artistic creation, emphasizing the lack of consent, credit, or compensation for original creators.
• Uses the metaphor of 'stochastic Pollocks' to describe AI systems that produce random, derivative outputs while obscuring the real labor of artists.
• Reviews public protests and artist-led movements (e.g., bans on AI art in creative platforms) to substantiate claims of systemic labor theft.
• Evaluates the legal and ethical frameworks governing AI training data, highlighting gaps in consent and attribution mechanisms.

Experimental results

Research questions

• RQ1 To what extent do generative AI models rely on unconsented, uncredited artistic labor in their training data?
• RQ2 How does the current AI art ecosystem enable the extraction of value from human artists without fair compensation?
• RQ3 In what ways does the creation of AI-generated art mirror or differ from traditional artistic labor in terms of ethical and economic value?
• RQ4 Why is the artistic community's framing of AI art as 'theft' ethically and structurally defensible?
• RQ5 What broader implications does AI art's labor extraction model have for other AI applications in creative and professional domains?

Key findings

• Generative AI models such as DALL-E, Midjourney, and Stable Diffusion are trained on massive datasets of copyrighted artworks collected without consent or compensation.
• The training process constitutes a form of labor theft, as human artists' creative efforts are extracted and repurposed without attribution or remuneration.
• The term 'stochastic Pollocks' effectively captures the paradox of AI art: algorithmically random outputs that mimic artistic creation while erasing the real labor of human creators.
• The artistic community's protest movement against AI art is grounded in a legitimate ethical claim, as AI systems systematically devalue and displace human artistic labor.
• The ethical failures in AI art are not isolated but indicative of a broader pattern of labor extraction in AI development, raising concerns for other AI applications.
• The paper concludes that many other AI applications likely rely on similar forms of unethical data extraction, suggesting a systemic issue in AI innovation.