Vivat Academia (2025).

ISSN: 1575-2844

Received: Apr 29, 2025 Accepted: Sep 1, 2025 Published: Sep 10, 2025

ARTIFICIAL INTELLIGENCE IN ARTISTIC CREATION: INNOVATION, HOMOGENEITY AND CHALLENGES OF ORIGINALITY

Virginia Puertollano Galán[1]: Francisco de Vitoria University. Spain.

Maribel Castro-Díaz : Francisco de Vitoria University. Spain.

mi.castro.prof@ufv.es

How to cite the article:

Puertollano Galán, Virginia & Castro-Díaz, Maribel. (2025). Artificial Intelligence in Artistic Creation: Innovation, Homogeneity, and Challenges of Originality. Vivat Academia, 158, 1-17. https://doi.org/10.15178/va.2025.158.e1608

ABSTRACT

Introduction: Artificial intelligence (AI) transforms artistic creation, allowing for rapid image generation while also posing challenges in terms of originality and visual diversity. This study analyzes the impact of AI on creativity, aesthetic homogenization, and the legal and ethical issues related to authorship. Methodology: A literature review was conducted on the use of AI in art, addressing technological, aesthetic, and legal aspects. Emerging regulatory frameworks were examined, along with studies on AI's reliance on pre-existing data. Results: While AI has democratized access to visual creation, its dependence on training data limits innovation and encourages stylistic repetition. A risk of aesthetic homogenization was identified, as well as legal gaps regarding the authorship of AI-generated works, since current regulations do not account for AI's role in artistic creation. Discussion: AI functions more as a tool for reinterpretation or data curation rather than as an autonomous creator. Its widespread use could lead to a standardized aesthetics, affecting artistic diversity. To mitigate these effects, strategies such as prompt engineering and the combination of digital and manual techniques are suggested, along with a regulatory framework to protect originality. Conclusions: AI expands creative possibilities, but poses challenges in terms of originality and authorship. A critical approach and collaboration between artists and technology are key to preserving aesthetic diversity and preventing the standardization of artistic production.

Keywords: Artificial intelligence, design, creativity, prompts, aesthetic homogenity.

1. INTRODUCTION

In recent years, powerful multimodal generative models have emerged. OpenAI, for example, released DALL·E 3 in 2023, as well as the Sora video model and GPT-4 with vision capabilities (and its variant, GPT-4o). These systems can generate complex images and videos from textual descriptions that are coherent and rich in expression in a way that has never been seen before. Tools such as Midjourney (updated in 2024), Stable Diffusion XL, and Adobe Firefly enable users to instantly create scenes in various styles, such as impressionist or surrealist, based on a brief prompt.

This democratizes digital artistic production, as users without technical training can create appealing digital images using only text. However, several recent studies emphasize that these models learn from vast amounts of visual data, making them more stylistic replicators than genuinely innovative generators. In other words, their creativity is based on recombining learned patterns rather than inventing entirely new styles. Nevertheless, artistic projects using AI, such as OpenAI's Sora Selects program, are exploring these systems' creative potential by showcasing experimental short films by AI-supported artists.

Artificial intelligence (AI) is redefining perceptions of art in the digital age by questioning traditional limits of visual expression and rethinking concepts of art and creativity (Vertedor Romero, 2023). Human beings are progressively granting greater authority and autonomy to technologies for artistic creation to the point that AI is considered an independent creative agent rather than a mere tool (de Propios Martínez, 2022).

This article is a theoretical-reflective essay and is not intended to be an applied, empirical study. Rather, it is a conceptual approach to the role of AI as a creative agent in artistic production.

2. OBJECTIVES

This research aims to address a series of key objectives and answer the posed questions by delving deeper into the subject of study.

1. To analyze how AI influences creative processes that use this technology to generate new aesthetics and styles.

2. To explore the challenges that AI models generate from large datasets and the dangers of aesthetic homogeneity.

3. To address issues of originality and authorship in AI-generated imagery to reflect on the degree of innovation in these works.

4. To propose strategies for creators to combat the uniformity of AI-generated works and develop personal, genuine styles.

3. METHODOLOGY

This research employed a methodology involving a review of academic sources and recent scientific studies on the implications of applying AI to artistic creation. Articles and academic publications by renowned authors, such as J. Martín Prada, were examined. Additionally, legal studies on issues of originality and rights protection in AI-generated works were reviewed (Mazzi, 2023).

Research on debates and threats to creating original works through AI received special attention, and the importance of prompt engineering in the creative process was supported by specific studies (Korzynski et al., 2023). Prompt engineering helps artists achieve personalized and specific results with greater autonomy.

In addition to the literature review, the impact of generative models on visual aesthetics was considered from the perspective of the creative experience. Finally, the legal implications of copyright were examined, including the potential adoption of neighboring rights in the European context, as discussed in studies of current debates surrounding the intellectual property of artistic works generated by AI (Kucukali, 2022).

4. RESULTS

4.1. The Limits of AI in Creative Innovation

Artificial intelligence (AI) has brought about a profound revolution with tools that allow for the near-instantaneous exploration of patterns, styles, and combinations. Generative adversarial networks (GANs), for example, have gained immense popularity for enabling the creation of images based on pre-existing styles (Jiang et al., 2022). Today, anyone can generate a Disney-style scene or a Picasso Cubist portrait, thanks to the vast amount of data available to AI.

However, AI's dependence on training data and the current architecture of these models limits its ability to produce something radically new. AI systems function more as stylistic replicators than autonomous generators of novel images, repeating patterns and models from the training dataset (Paquette, 2021).

Furthermore, AI faces a fundamental challenge: its capacity for innovation is proportional to the data it consumes. By relying on large databases of historical and contemporary artworks, AI models perpetuate the aesthetics and themes that dominate those sources. This can restrict the emergence of disruptive styles. Thus, AI's role may be more as a tool for reinterpretation than pure creation. This perception may lead to viewing AI as a "curator" or "editor" of pre-existing influences rather than an independent creator (Mazzi, 2023).

4.1.1. AI as a Tool for Reinterpretation and Generation of Variations

Generative AI models, particularly generative adversarial networks (GANs), allow for the quick and systematic reinterpretation of patterns and styles. Consequently, artists and creators can use AI to produce styles and aesthetics in seconds that previously required considerable effort. This allows them to explore diverse variations of a visual or auditory idea without committing their time or dedication to a single concept. Tools such as DALL-E and Midjourney allow users to transform a photographic image into an Impressionist or Surrealist style, for example, and generate final artwork without resorting to sophisticated manual techniques. This contributes to the democratization of digital artistic production because it enables anyone, regardless of their technical capabilities or artistic skills, to create appealing digital images based on simple textual descriptions.

However, AI's ability to immediately reinterpret images calls into question the type of creativity it fosters. Often, AI functions as a stylistic filter that transforms the input provided by the creator. Rather than producing something entirely new, AI adjusts the input information and the material provided. This limits creation to the combination or variation of pre-existing styles. While this approach can be useful for visualization and stylistic experimentation, it is not suitable for radical creative innovation because it is determined by defined structures and patterns.

4.1.2. Limitations of Generative Models in Aesthetic Innovation

Although GANs and other AI systems are powerful tools for creating stylistic variations, their capacity for radical innovation is limited. Since they are trained on existing data, they tend to replicate and recombine patterns rather than invent entirely new paradigms (Paquette, 2021). Even the most recent models (2023–2025), which have improved semantic coherence and multimodal generation (text, image, audio, and video), continue to operate based on statistical correlations rather than lived experiences or cultural contexts. Juan Martín Prada highlights this ambivalence. While these tools enable the rapid and inexpensive creation of images, their reliance on existing images can seriously devalue human creativity in many dimensions (Martín Prada, 2024).

Although this stylistic repetition is often considered a limitation of AI, it is worth remembering that human artists also repeat themselves stylistically, like Picasso or Beethoven. The difference is that human repetition occurs within a historical and personal process of expressive exploration, whereas algorithmic repetition responds to statistical recombination of training data (Martín Prada, 2024; Chang et al., 2023).

4.1.3. Dependence on Training Data and Pre-existing Trends

An AI's capacity to innovate is fundamentally limited by the quality and diversity of its training data. AI models, especially generative adversarial networks (GANs), are trained using large amounts of visual, auditory, or textual data representing a wide range of artistic styles and themes. However, this dependence on training data implies that AI inevitably perpetuates the trends and biases present in the data.

For instance, if a model is trained primarily with data on Western art, its creations will likely reflect the characteristics of that cultural context. This limits diversity and perpetuates pre-existing aesthetic norms, as Mazzi (2023) has argued. This is particularly relevant in a global artistic context where styles and cultural representations have unique, defining value. By extracting only what is represented in its training data, AI can restrict the cultural and stylistic spectrum of creativity. Furthermore, as Kucukali (2022) notes, critics argue that this can lead to AI tools contributing to the "standardization" or "globalization" of aesthetics. In this scenario, AI-produced works become interchangeable, eliminating the unique differences that traditionally characterize diverse forms of cultural art.

In the face of increasing standardization, it is important to remember that human creativity is diverse and depends on the creators and their contexts. However, AI will develop in more diverse ways, taking into account more contexts and values (Sáez-Velasco et al., 2024).

4.1.4. The Role of the Human Creator in Collaborative Innovation with AI

Collaborative Innovation with AI

The intervention of the human creator is essential for overcoming some of the aforementioned limitations and introducing a sense of individuality into the artistic creation process with AI. Techniques such as prompt engineering allow artists to influence AI models to generate works that align with their visions and aesthetic preferences. Prompt engineering enables artists to establish specific parameters, allowing the model to generate results that best reflect their personal style. However, this process ultimately depends on the capabilities of the model.

Christian Vinces coined the term "promptography," which involves using specific prompts or commands to generate images. In this process, the interaction between natural language and generative models is important (Vertedor Romero, 2023). The image generation process is not entirely predictable. The algorithms used are complex; therefore, there is an element of surprise in the result. For this reason, the process is usually iterative. The human creator generates variations of the visual results offered until selecting the one that best aligns with their intentions (Martín Prada, 2024). Ultimately, users must know how to ask the AI for what they want to truly obtain a conceptually and visually satisfying image that effectively embodies their artistic vision.

The collaboration between humans and AI introduces a new concept of creativity in which AI acts as an extension of the artist's intentions and vision. AI's true value lies in its ability to assist and amplify human creativity rather than replace it entirely (Korzynski et al., 2023). This collaboration redefines authorship in art, where the artist is both the creator and the "director" of an AI operating under their guidance. This approach transforms the artist's role into that of a "curator of styles and patterns," with AI becoming a new method for bringing their visions to the digital realm, allowing them to explore complex ideas more quickly and accurately.

For this collaboration to be effective, however, artists must adopt a critical attitude toward AI. They must understand its limitations and seek new ways to intervene manually or with hybrid methods to maintain the individuality of their works (Kucukali, 2022).

In this new context of artificial intelligence, creative innovation depends as much on the technical capacity of generative models as on artists' ability to interact with and manipulate them. While AI provides numerous possibilities for exploring styles and patterns, its ability to innovate is constrained by its dependence on training data, databases, and AI structure. Human collaboration is therefore fundamental to avoid aesthetic homogeneity and to foster innovation that is not only technical but also conceptual and cultural. As Martín Prada states, “what is fundamental in art is always a happy combination between idea and realization, or, in Kantian terms, that suitability between two mental faculties: imagination and understanding” (2024, p. 5).

4.2. Aesthetic Homogeneity: A Risk to Diversity in Art

One of the most significant risks of AI in art is the creation of a homogeneous aesthetic. While recent models such as DALL·E 3, Stable Diffusion XL, Midjourney (2024), and Adobe Firefly have expanded technical capabilities, they have also accentuated this tendency toward uniformity. Generative adversarial networks (GANs) and other generative models tend to produce visually appealing images that often share a similar "patina" or visual style, a product of common patterns in the training data. As authors such as Paquette (2021) and Mazzi (2023) have explained, the result is a series of works that, though varied in detail, appear to conform to a single style. This limits aesthetic diversity.

This phenomenon of aesthetic homogeneity is problematic for art, which has historically been enriched by diverse expressions and paradigm shifts. Uniformity in AI-generated art could lead to a kind of stylistic "saturation," where viewers are confronted with AI works that, while initially impressive, have a generic appearance and lack the freshness and originality associated with human artistic creation. Critics suggest that, rather than innovating, AI could contribute to the "mechanization" of creativity, where stylistic variation is superficial rather than profound (Korzynski et al., 2023). Martín Prada (2024) speaks of "parasitic creativity" because AI imitates images and styles developed by humans. Ultimately, AI also imitates images created by AI itself.

To address this issue, it is recommended that artists adopt prompt engineering strategies to guide AI models toward personalized results that align with their unique aesthetic preferences and creative vision. Prompt engineering provides greater control over the creative process, reducing aesthetic homogeneity and enabling artists to maintain a unique voice in their work. However, this approach has significant limitations. The range of variation ultimately depends on the models' capabilities and the underlying data (Jiang et al., 2022). Some corporations, such as Adobe, are exploring ways to enable artists to train machine learning models to generate content reflecting their personal style without external influence (Martín Prada, 2024).

4.2.1. The Effect of Homogeneity on the Art Market and Consumption

The growing popularity of AI-generated artworks and their unique aesthetic has begun to influence market expectations. Between 2023 and 2025, NFT platforms, digital galleries, and social media show a growing demand for the so-called "AI aesthetic" (Chi, 2024), which reinforces the trend toward uniformity. Since AI-generated art often shares certain visual characteristics, there is now a specific demand for this type of digital aesthetic. In some cases, this demand presents itself as a uniform trend across online platforms and galleries. This demand may pressure artists to conform to the general market's taste by following "trendy" AI-generated styles, which could limit diversity in the production of new styles and creative approaches.

Furthermore, the market's normalization of AI-generated art could overshadow more traditional or manual forms of artistic creation. This leads to standardized art consumption, where the characteristics of AI-generated works are prioritized over the creative diversity of original pieces. Consequently, standardization affects not only production but also cultural consumption, favoring dominant visual languages and limiting the visibility of alternative proposals.

4.2.2. The Visual "Patina" as a Result of the Data Feedback Loop in Content Creation

Due to similarities in data processing and learned patterns, GANs and other AI models tend to create works that share a homogeneous visual appearance. This "patina" is the result of the influence of the training data and the algorithmic structure of the model. As AI models replicate patterns, a lack of visual diversity becomes apparent. The uniformity of color palettes, textures, and styles can lead viewers to see a series of images that, although detailed and complex, appear similar and predictable aesthetically.

This visual homogeneity stems from the fact that, while powerful, AI models lack the contextual or cultural awareness necessary to develop a personalized or radically innovative aesthetic. AI simply reconfigures existing patterns, resulting in outputs with a cohesive yet limited appearance. The phenomenon of "patina" limits not only the perception of AI-generated art but also presents a risk of cultural standardization, where certain styles dominate over others due to their replicability by AI, as authors like Mazzi (2023) have pointed out. De Propios Martínez (2022) has noted that, in the process of learning from human-produced data, artificial intelligence systems exhibit flaws related to the inaccuracy of the information. These biased machine learning systems lead to predictable results.

A feedback loop in the training data reinforces the aesthetic homogeneity in AI-generated art. When AI models are trained with large volumes of visual data, they tend to learn the predominant patterns and aesthetics of the available content. This process is amplified when AIs are trained with works previously generated by other models. This creates a cycle in which the AI learns and constantly reproduces a digital aesthetic similar to those that already exist. This limits the diversity of works produced and reinforces a self-referential aesthetic.

Most AI models have their own default aesthetic that is applied unless instructed otherwise. As more artists and creators use AI to generate their work, those works can become part of future datasets, thereby reinforcing existing styles and patterns. Ultimately, AI replicates and amplifies its own stylistic limitations, closing the door to diversity and originality in future creations.

4.2.3. Strategies to Mitigate Aesthetic Homogeneity

Despite the novel challenges observed in this study, there are several resources artists can use to combat potential visual uniformity in AI-assisted works. One interesting strategy is prompt engineering, which enables creators to direct the models' output more personally. By providing detailed descriptions and instructions, artists can direct the AI toward results that reflect their personal vision, adding an element of uniqueness and personalization to their work.

Another strategy is to combine manual and digital methods so the AI acts as a starting point for creative experimentation and manual editing techniques are applied to add authenticity and distinction. This hybrid approach strengthens the artist's role as an active participant in the creative process. Furthermore, using artificial intelligence does not exempt creators from having basic knowledge, such as color theory, narrative, and composition. These remain essential for understanding and interpreting the created content (Sáenz-Velasco et al., 2024).

Additionally, diversifying the training data for the models to encompass a wider range of styles and visual cultures can reduce the likelihood of a homogenous aesthetic and broaden the creative spectrum (Jiang et al., 2022). Human intervention and strategic planning are essential to ensure that art created by AI is not a standardized product but a genuine, varied reflection of artistic creativity. Along these lines, incorporating AI technologies into art education and promoting their study and development in cultural institutions has been emphasized (de Propios Martínez, 2022).

Although the debate has largely centered on Western contexts, innovative proposals are emerging in other environments as well (Herrera Ávila, 2025; Manovich & Arielli, 2024; Afegbua, 2023). In Latin America, for example, studies conducted in Mexico demonstrate the incorporation of AI into collaborative practices and mythological narratives. In South Korea, for example, artists like L. Manovich use models trained in the Korean language to adapt digital aesthetics to their local culture. In Nigeria, creator Malik Afegbua developed The Elders Series (2023), which uses artificial intelligence to challenge stereotypes of old age. These examples demonstrate that the creative use of AI is not exclusive to the West but rather interacts with diverse cultural contexts.

4.3. Originality and Authorship in AI-Generated Art

Another major debate in AI-generated art concerns originality and authorship. AI challenges the traditional notion of authorship, in which the human creator is considered the primary source of creativity and originality. In the case of generative systems, for example, the human creator could be seen as a facilitator or editor, while the AI model acts as a co-author in the creative process. This collaboration raises questions about credit and authenticity in AI-generated works, as well as who should receive the copyright for these creations (Paquette, 2021).

The concept of originality in the context of AI is highly complex and subject to extensive debate. While an AI-generated work may be technically original if it differs from the images in its dataset, it often lacks the emotional depth and cultural context associated with the creative process of artists and creative professionals. Current copyright regulations in jurisdictions such as the European Union do not address works generated autonomously by AI. However, approaches based on neighboring rights have been proposed that would recognize the technology's contribution without supplanting human authorship (Mazzi, 2023).

Authors like Paquette (2021) and Hristov (2016) argue that AI-generated works should be in the public domain because they lack clear, direct human contribution and do not meet the traditional originality criteria requiring human intervention. However, some argue that the copyright framework should be adapted to allow for a protection regime that recognizes the role of AI while still protecting the rights of human creators. This could be achieved through contracts or licenses that stipulate collaboration between humans and AI in artistic creation.

Currently, creators cannot prevent their works from being included in the databases used by many AI models for training purposes. This poses a threat to creators given the possibility that AI models will generate "new" images by imitating their style. As Martín Prada (2024) points out, current copyright legislation only protects the work itself, not the style. This ignores the fact that the core of contemporary artistic creativity is the creation of a unique style. This is something that AI itself is not yet generating in an innovative way.

4.3.1. The Issue of Authorship: AI as Creator or Tool

The concept of authorship in AI-generated art is central to discussions about intellectual property rights. Traditionally, authorship is attributed to the person who contributes creativity and intentionality to a work. However, AI does not possess these attributes autonomously. Thus, there is debate over whether AI should be recognized as a creator or as an advanced tool that assists the true human creator.

According to some studies, such as Jiang et al. (2022), AI performs most of the visual generation work, but human intervention through prompt engineering or subsequent image manipulation is essential to the creative process. In this sense, the artist functions as a director, establishing the aesthetic and conceptual intention. This reinforces the idea that the ultimate author should be human, even if AI contributes significantly to the technical aspects of the result.

These technologies offer human creators an enormous capacity to visualize images simply by describing them. Tools like Midjourney and DALL·E 3 can unlock anyone's creativity and allow them to experiment with visualizing ideas. However, as Korzynski et al. (2023) point out, this democratization can result in a loss of authorial singularity if the results differ little from common algorithmic aesthetics.

4.3.2. Copyright and Emerging Legal Frameworks

The current copyright system was not designed to address AI-generated creations, creating a series of legal and regulatory challenges. In many jurisdictions, copyright protection is linked to human intervention, excluding works generated autonomously by machines. The rapid adoption of AI in artistic creation, however, has prompted regulatory bodies such as the European Union to consider adjustments to their legal frameworks (Kucukali, 2022).

In 2024, the European Union proposed regulatory adjustments through the AI Act and complementary intellectual property directives, including recognizing neighboring rights for works partially generated by AI (European Commission, 2025).

One solution proposed at the European level is establishing a neighboring rights regime that would grant certain rights to producers or users of AI systems without considering the AI as the direct author. This would provide partial protection and prevent AI-generated works from automatically entering the public domain. However, this approach raises complex debates about the limits of protection and the fair distribution of rights, especially in scenarios where the roles of AI and the human creator are difficult to distinguish. Meanwhile, the US Copyright Office (2023–2024) has issued resolutions denying copyright to works generated without significant human intervention. These resolutions reaffirm that protected creativity remains exclusively human (Gourvitz & Ameri, 2023; Samuelson, 2025).

4.4. Ethical Perspectives on Authorship and Credit in Collaborative Creation

AI-generated art raises questions about authorship and challenges the nature of this artistic genre (de Propios Martínez, 2022). Beyond legal issues, there are ethical implications related to recognizing and attributing credit to AI-generated works. Traditionally, art creation has been viewed as a profoundly personal and human activity. In this context, authorship implies a degree of emotional and conceptual commitment that AI cannot replicate.

De Propios Martínez (2022) argues that AI techniques are often considered independent creative agents. However, relationships regarding authorship are often contradictory. As Hristov (2017) points out, recognizing AI as a co-author could trivialize the value of human labor and blur the lines between creative intervention and technical execution. Furthermore, failing to recognize the human contribution to the AI generation process could lead to the devaluation of the artist's role in favor of excessive dependence on technology. This could result in a model of artistic production where human creators are relegated to a secondary role, or worse, eliminated from the creative process altogether. Therefore, the ethical challenge lies in finding a balance that recognizes the value of collaboration without displacing the human artist as the central figure in creation.

Regarding the use of artists' works as a basis for training generative models, Martín Prada (2024, p. 8) insists that:

(…) many artists consider this a form of non-consensual expropriation of their work because their creations are used to develop new images without considering any of the three "C's" that form the core of their demands: consent, credit, and compensation.

Originality and authorship in AI-generated art present two interrelated and complex challenges. The ambiguity in defining originality in the context of AI, coupled with the difficulty of establishing clear authorship, underscores the need for a legal and ethical framework that provides adequate protection without trivializing the human contribution. Potential solutions include adjustments to the current copyright system and creating related rights for AI-generated works. At the same time, human intervention in the creative process must be recognized as an essential component that preserves the value of artistic creation amid increasing technological dependence.

5. CONCLUSIONS

AI offers expansive possibilities and unique challenges to contemporary art. Its reliance on pre-existing data can limit innovation and foster homogeneity, which poses a risk to creative diversity.

However, techniques such as prompt engineering and manual intervention enable artists to exercise greater control over the creative process and its results.

The question of authorship in AI-generated art remains unanswered, underscoring the necessity of a legal and ethical framework that addresses collaboration between humans and machines. Recognizing the rights of both human creators and AI systems through a flexible approach could be the next step in the evolution of art and intellectual property in the digital age.

Originality in AI-generated art is limited by the technology's algorithmic nature and dependence on training data. Although generative adversarial networks and other AI models can produce visually complex works, these pieces tend to reflect patterns and styles already present in the models' training data. This suggests that, while they are technically original in that they are new creations, they may not be truly original. AI-generated works often lack the conceptual depth and disruptive capacity that have characterized innovations throughout art history. Artists also develop human-produced works with stylistic repetitions, as seen in musicians like Beethoven or painters like Picasso. However, human repetition is not mere copying; it is a progressive exploration that yields different meanings with each iteration. In AI, repetition arises from algorithmic calculations on pre-existing data, limiting its capacity to generate genuine conceptual breakthroughs.

Another important finding is the presence of a shared visual "patina" in many AI-generated works. This limits aesthetic diversity and poses a challenge in an artistic context that is characterized by heterogeneity and the ability to break molds. AI's standardization could lead to the appropriation and saturation of certain styles, particularly on digital platforms where AI-generated art is widely consumed and reproduced. These findings suggest the need for advanced techniques, such as prompt engineering and diversification of training data, to mitigate the effects of aesthetic homogeneity and preserve individuality in art.

AI also challenges traditional notions of authorship in art by blurring the line between human creator and machine. Although human artists provide prompts and establish aesthetic direction, AI models are responsible for executing the creative process (Kasap, 2019). This raises dilemmas about authorship and copyright.

On a positive note, AI has opened new possibilities for creative collaboration, enabling artists to leverage technology as an extension of their creative process. Artists can maintain their creative voice while exploring digital aesthetics through prompt engineering and manual intervention techniques. As Vertedor Romero (2023) points out, exploring new forms of artistic expression and reflecting on the connection between technology and the human essence will be fundamental to 21^st-century art. The accelerated evolution of AI up to 2025, with increasingly sophisticated and multimodal systems, confirms that the debate on originality, homogeneity, and authorship is not static, but rather requires constant updates. This work aims to offer a theoretical framework that serves as a reference point amid continuously evolving technologies.

The results underscore the necessity of establishing an ethical and regulatory framework that addresses the implications of AI-generated art and protects the roles of human creators and artistic diversity. Emerging policies and regulations, such as neighboring rights in the European Union, represent a preliminary effort to protect AI-generated art.

Additionally, it is crucial to expand research to non-Western contexts to determine if the homogenization and authorship patterns observed in European and North American environments are replicated in other cultural traditions or if unique creative models emerge that enrich the global digital art landscape.

Overall, AI has the potential to enrich the creative process, though it raises questions about originality, aesthetic homogeneity, and authorship. Human-machine collaboration is a viable way to maintain artistic individuality and requires a regulatory framework balancing technological interests and human creative rights.

6. REFERENCES

Afegbua, M. (2003). Inclusive Digital Futures. https://www.malikafegbua.com/

Chang, M., Druga, S., Fiannaca, A., Vergani, P., Kulkarni, C., Cai, C., & Terry, M. (2023). The Prompt Artists. En Proceedings of the 15th Conference on Creativity and Cognition (pp. 75-87). Association for Computing Machinery. https://doi.org/10.1145/3591196.3593515

Chi, J. (2024). The evolutionary impact of artificial intelligence on contemporary artistic practices. Communications in Humanities Research, 35(1), 6–11. https://doi.org/10.54254/2753-7064/35/20240006

European Commission. (2025). AI Act. En Shaping Europe’s digital future. https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai

Gourvitz, E., & Ameri, L. (2023). Can works created with AI be copyrighted? Copyright Office issues formal guidance. Ropes & Gray. https://www.ropesgray.com/en/insights/alerts/2023/03/can-works-created-with-ai-be-copyrighted-copyright-office-issues-formal-guidance

Herrera Ávila, T. (2025). Ficciones algorítmicas: Desafíos éticos y políticos de la era de la inteligencia artificial. Astrolabio. Revista Internacional de Filosofía, 30, 1-12. https://doi.org/10.1344/astrolabio.v1i30.50297

Hristov, K. (2017). Artificial Intelligence and the Copyright Dilemma. IDEA: The Journal of the Franklin Pierce Center for Intellectual Property, 57(3), 431-454. https://ssrn.com/abstract=2976428

Jiang, Y., Li, X., Luo, H., & Yin, S. (2022). Quo Vadis Artificial Intelligence? Discover Artificial Intelligence, 2(4). https://doi.org/10.1007/s44163-022-00022-8

Kasap, A. (2019). Copyright and Creative Artificial Intelligence (AI) Systems: A Twenty-First Century Approach to Authorship of AI-Generated Works in the United States. Wake Forest Journal of Business and Intellectual Property Law, 19(4), 335-380. https://doi.org/10.31235/osf.io/gnyha

Korzynski, P., Mazurek, G., Krzypkowska, P., & Kurasinski, A. (2023). Artificial intelligence prompt engineering as a new digital competence. Entrepreneurial Business and Economics Review, 11(3), 25-37. https://doi.org/10.15678/EBER.2023.110302

Kucukali, B. T. (2022). The Protection of AI-Generated Works under European Copyright Law: Toward Adoption of a Neighbouring Rights Approach. University of British Columbia. https://dx.doi.org/10.14288/1.0421298

Manovich, L., & Arielli, E. (2024). Artificial aesthetics: Generative AI, art and visual media. https://manovich.net/index.php/projects/artificial-aesthetics

Martín Prada J. (2024). La creación artística visual frente a los retos de la inteligencia artificial. Automatización creativa y cuestionamientos éticos. Eikón / Imago, 13, e90081. https://doi.org/10.5209/eiko.90081

Mazzi, F. (2023). Authorship in artificial intelligence-generated works: Exploring originality in text prompts and artificial intelligence outputs through philosophical foundations of copyright and collage protection. The Journal of World Intellectual Property,27, 410-427. https://doi.org/10.1111/jwip.12310

Paquette, L. (2021). Artificial Life Imitating Art Imitating Life: Copyright Ownership in AI-Generated Works. Intellectual Property Journal, 33, 183-215. https://www.proquest.com/openview/bd7c4eeb0501273765a62eca437241c0/1.pdf?cbl=46743&pq-origsite=gscholar&utm_source=chatgpt.com

de Propios Martínez, C. (2022). Arte e inteligencia artificial: técnicas de aprendizaje automático en el arte generativo actual [Tesis doctoral]. Universidad Complutense de Madrid. https://hdl.handle.net/20.500.14352/3923

Sáez-Velasco, S., Alaguero-Rodríguez, M., Delgado-Benito, V., & Rodríguez-Cano, S. (2024). Analysing the Impact of Generative AI in Arts Education: A Cross-Disciplinary Perspective of Educators and Students in Higher Education. Informatics, 11(2), 37. https://doi.org/10.3390/informatics11020037

Samuelson, P. (2025). AI authorship revisited. Assessing the evolving perspectives regarding the copyright aspects of generative artificial intelligence outputs. Communications of the ACM, 68(2), 26-28. https://doi.org/10.1145/3732797

Vertedor Romero, J. (2023). Editorial. AI-driven art: la inteligencia artificial en el arte y el diseño. UMÁTICA. Revistas sobre creación y análisis de la imagen, 5(6), 9-20. https://doi.org/10.24310/umatica.2023.v5i6.18315

AUTHORS' CONTRIBUTIONS, FUNDING AND ACKNOWLEDGMENTS

Authors' contributions :

Conceptualization: Puertollano Galán, Virginia and Castro Díaz, Maribel. Methodology: Puertollano Galán, Virginia and Castro Díaz, Maribel. Software: Puertollano Galán, Virginia and Castro Díaz, Maribel. Validation: Puertollano Galán, Virginia and Castro Díaz, Maribel. Formal Analysis: Puertollano Galán, Virginia and Castro Díaz, Maribel. Data Curation: Puertollano Galán, Virginia and Castro Díaz, Maribel. Drafting and Preparation of the Original Draft: Puertollano Galán, Virginia and Castro Díaz, Maribel. Drafting, Review, and Editing: Puertollano Galán, Virginia and Castro Díaz, Maribel. Visualization: Puertollano Galán, Virginia and Castro Díaz, Maribel. Supervision: Puertollano Galán, Virginia and Castro Díaz, Maribel. Project Management: Puertollano Galán, Virginia and Castro Díaz, Maribel. All authors have read and accepted the published version of the manuscript: Puertollano Galán, Virginia and Castro Díaz, Maribel.

Funding: This research did not receive external funding.

AUTHOR(S):

Virginia Puertollano Galán: Professor at Francisco de Vitoria University, teaching in the Fine Arts and Design programs. She holds a Bachelor of Fine Arts and a diploma in real-time 3D applications from Francisco de Vitoria University. She also holds a master's degree in art and design of video games from U-Tad and a master's degree in AI and innovation from Founderz. She specializes in 3D modeling and she is a member of the Innovation and Image Analysis (IAI) Research Group. She participates as a researcher in the ANDROMEDA project: Myth and Representation: Theoretical and Practical Activities of Innovation in Cultural Myth Criticism" (Community of Madrid, PHS-2024/PH-HUM-76).

virginia.puertollano@ufv.es

Orcid ID: https://orcid.org/0009-0007-6950-2817

Google Scholar: https://scholar.google.es/citations?user=feWfzxoAAAAJ&hl=es

Academia.edu: https://ufvitoria.academia.edu/VirginiaPuertollano

Maribel Castro-Díaz: Associate Professor at Francisco de Vitoria University, teaching in the Fine Arts and Design programs. She holds a Bachelor of Fine Arts degree from the University of Vigo (with an Extraordinary Award) and a doctorate in Fine Arts from the Complutense University of Madrid (with an Extraordinary Doctoral Award). She is accredited by ANECA as a full professor and has a six-year research period. She belongs to the Innovation and Image Analysis (IAI) Research Group and is a researcher on the ANDROMEDA project. Myth and Representation: Theoretical and Practical Activities of Innovation in Cultural Myth Criticism" (Community of Madrid, PHS-2024/PH-HUM-76).