This essay presents a series of reflections developed as part of an on-going artistic research umbrella project titled “The Unexpected.” The project explores human-AI interactions, touching on topics such as cognition, emotions, feelings, artificial intelligence, and everything between these fields.

This interdisciplinary PEEK/FWF-funded project brings together artists Pamela Breda, Patricia Reis and Ruth Schnell (University of Applied Arts Vienna), and researchers Matthew Pelowski (University of Vienna), Rebecca Chamberlain (Goldsmiths University), and Catherine Weasley (King’s College). The goal is to study how humans approach AI software in perceptive situations where the lines between reality and fiction become increasingly blurred – from social networking to deepfake videos to humanoid-looking robots – and the feelings and thoughts such interactions generate.1

Moving on from this broader research framework, my essay explores the multifaceted aspects of CGI, considering its historical evolution, its profound influence on various domains of individual and collective life, and the philosophical quandaries it introduces. In particular, the analysis will consider how hyperrealistic CGI reshapes our perception of reality, confronts the age-old debate between authenticity and artifice, and redefines the boundaries of human creativity in the digital age.

Computer-generated images, also known as CGI, have a relatively short history compared to other forms of visual art. Their development can be traced back to the early 1960s, when computer scientist Ivan Sutherland developed a program called Sketchpad, the first system to enable users to create graphical images directly on a computer screen. Sketchpad was a breakthrough because it allowed users to manipulate objects in three dimensions. Such a technology had never been available to the general public before.

In the 1970s, with advancements in computer technology, CGI began to be employed in films. A famous example is the 1973 movie “Westworld,” which featured a computer-generated hand. The first fully computer-generated film was “Toy Story,” which was released in 1995. Produced by Pixar Animation Studios, the film used a technique called 3D rendering to create realistic, computer-generated characters and environments.

Since then, computer-generated images have become a staple of the film industry, with many movies using CGI to create everything from characters to entire worlds. In the last decades, CGI technology has also been used in video games, architecture, engineering, and many other fields. One of the most recent advancements in this domain is DALL·E, a neural network-based image generation model developed by OpenAI and released in 2020 (Bobby, 2022).2

From a simple “script” – a textual description of a specific image – DALL·E can create hyperrealistic pictures of objects, people, and events that do not exist in the real world. The software is trained on a massive dataset of text and visual material to learn how to generate realistic and coherent images and has become a popular tool for artists, designers, and creators. However, it has also sparked controversies on copyright issues since some artists sued the company for using their images for training purposes without permission (Ivanova 2023), as well as a more general concern about the role of AI in substituting human creativity.

Notably, critics like Lev Manovich (2001) and Nora Khan (2019) have extensively explored how artificial intelligence fundamentally alters the landscape of visual culture. In studying the language of new media such as digital arts, computer games and social media, Manovich posited that a software is not just a tool, but a cultural form in its own right (Manovich 2001, p. 194). His theory focuses on the concept of “database logic,” referring to the way in which digital media is organized and structured.

According to this view, the database has assumed a pivotal role in shaping contemporary culture since it is not solely a means to an end but rather a dynamic component that actively shapes and influences the experiences and interactions of users within the digital realm.

In computer games, software is not merely a tool for executing code. It is an essential part of the gaming experience since it defines the rules, physics, graphics, and interactivity of the virtual worlds that players inhabit. In the realm of visual arts, software features, algorithms, and interfaces can profoundly impact the style, aesthetics, and communicative power of digital works of art. In the context of social media, software governs the structure and dynamics of online interaction. The algorithms that determine the content that users see and their engagement with it have significant societal and cultural implications. They shape what information people are exposed to, impacting their beliefs and worldviews, and incentivize certain types of content creation and online behaviors, which potentially reinforces echo chambers, political polarization, or other cultural dynamics. They can also raise questions about privacy, surveillance, and how the attention economy impacts individual and collective values.

Continuing beyond Manovich’s theories, Nora N. Khan (2019) examined the influence of artificial intelligence on the creative industries, such as generative art, computer vision, and the ethics of AI. This includes questions of misinformation and deception since AI can be used to create deepfake videos or images that deceive consumers. Additional ethical concerns include identity theft, the infringement of intellectual property rights, and privacy violations. Using hyperrealistic AI to generate realistic images or videos of individuals without their consent can lead to unauthorized surveillance, harassment, or the creation of inappropriate content.

Khan developed the concept of ‘aesthetics of the algorithm’, to exemplify how algorithms and machine learning systems are transforming how we experience art. According to Khan, algorithms are not simply tools that artists can use to create new works. They are increasingly shaping the very form and content of art.

The complex algorithms and data sets behind AI-generated art, however, are often opaque and difficult to understand, making it complex to evaluate the authenticity or originality of the work. AI systems often rely on vast datasets that process existing content. Potentially then, AI-generated creations could inadvertently replicate existing works, raising questions about intellectual property rights, originality, and plagiarism.

As we increasingly interact with AI that mimics human emotions (such as popular AI influencer Lil Miquela), engage in immersive virtual worlds mirroring our own (as in “The Last Of Us” videogame world), and consume content indistinguishable from reality (such as in the case of deepfake videos), we are forced to confront the question of what it truly means to be “real.”

The algorithms behind computational creativity – often inspired by human creative processes, such as analogy, exploration, and serendipity – rely on various techniques, including machine learning and generative models. In this context, AI engineers are actively investigating Embodied Intelligence theory, which proposes that intelligence is not solely governed by the brain, but is also influenced by the body’s interactions with the environment (Chrisley 2003). Cognitive scientist Rodney Brooks, one of the primary proponents of this theory, has argued that traditional approaches to artificial intelligence, based on symbolic processing and explicit knowledge representation, are insufficient for creating truly intelligent machines (Brooks 2001 and Brooks 2013). Instead, he proposed that intelligence emerges from the interaction between an agent’s body and its environment, without the need for explicit representations or reasoning.

Other researchers have attempted to identify specific neural correlates of consciousness in the human brain, to better understand how consciousness arises. In this regard, neuroscientist Giulio Tononi (2012) proposed an Integrated Information Theory, stating that consciousness arises when a system is capable of integrating a large amount of information in a highly structured and specific way. Additionally, researchers David Chalmers (2010) and Stanislas Dehaene (2014) have been exploring the Attention-Based Consciousness theory, which posits that attention is a critical factor in shaping consciousness, i.e., an artificial system could be deemed conscious if it possesses the ability to focus its attention on specific elements within its environment.3 This theory raises interesting possibilities when we consider its connection to the realm of AI-generated images. In this context, creativity often hinges on the system’s ability to focus its attention on certain elements within its dataset or creative space. Just as human consciousness involves the selective allocation of attention to different aspects of our environment, AI systems can emulate this process, using it as a foundational element for generating original content and artworks.

In the early 1970s, when computer-generated images first appeared, they were characterized by rudimentary attempts to replicate human creativity. Their visual content lacked the depth of human consciousness, experiences, and emotions.

However, contemporary hyperrealistic AI-generated images exhibit a notable departure from these earlier instances of CGI. The high degree of detail, texture, and realism achieved by AI technology has the power to create pictures that can deceive viewers as never before. What is even more interesting is that generative software is not merely replicating established styles or techniques but producing original content on the basis of specific prompts or instructions. This phenomenon not only reshapes conventional notions of art but also, redefines the concept of the human agent as a creative force capable of birthing a unique visual outcome.

Who should be credited as the author when the creative process involves both human input (programming AI) and machine autonomy (AI generating the art)? On one hand, the programmers and developers who craft the AI system undeniably play an instrumental role in shaping its abilities. They design the algorithms, select the training data, and fine-tune the model to create a machine that can generate unique visual content. In this sense, they serve as the architects and curators of the AI’s creative potential. As such, it seems reasonable to credit them for the initial design and programming, akin to the way an artist might be recognized for creating a unique brush or tool. However, if AI can replicate reality with incredible precision, does this diminish the creative act or redefine creativity to include the capacity to harness and manipulate algorithms and data? Does AI possess a form of creativity or is it merely a tool for human intention?

In this situation, I believe giving appropriate credit to both the software and the artists using it is crucial. At this current point in history, however, I believe we should regard the software as a medium through which artists express their creativity. It offers the tools, algorithms, and capabilities that help to craft unique artworks. Human creators then imbue these pieces with their distinct identity and emotional depth. Whether AI’s will be increasingly involved in generating original artworks in the future remains uncertain, however.

Traditionally, philosophers have distinguished between two approaches to representation: mimetic and constructive. In the mimetic approach, an image is considered successful when it accurately reflects or imitates the object it represents. Conversely, the constructive approach asserts that an image is effective when it can generate a novel object or distinct reality apart from the one it represents.

Hyperrealistic images blur the line between these two approaches as they are both highly mimetic (in their level of detail and accuracy) and highly constructive (in their creation of a new reality). CGI uses advanced 3D modeling, rendering, and animation techniques to create images, videos, and animations that are highly realistic, often to the point of appearing almost indistinguishable from photographs or live-action footage. This raises the question of whether hyperrealistic images are really “representations” at all, or whether they should be considered a new kind of object altogether.

This discussion echoes a historical parallel. When photography was first introduced, it challenged the traditional definition of art as representing reality. Critics debated whether a photograph, as a direct capture of a scene, could truly be considered art as it appeared to eliminate the need for an artist’s intervention in the creative process. Over time, it became evident that photography was indeed an art form in its own right, as photographers used various techniques to inject their unique vision and creativity into their work. Similarly, in the realm of CGI, artists employ their skills and artistic sensibilities to create images from a digital template. These are new kind of images, however, which require an in-depth analysis of their ontological status.

In this regard, we can move forward to the concept of ‘simulacra’, which was famously developed by Jean Baudrillard (1981).4 According to the French critic, simulacra are copies or representations that are detached from any original reality, leading to a situation where they become more “real” than the original itself. Are these hyperrealistic images mere representations born from script, or do they possess a reality of their own, independent of the original subjects they intend to represent? What are the real-life implications such images can have beyond the digital screen?

Let us examine Metahuman, a software created by the company Epic Games, which leverages AI technology to produce incredibly lifelike and customizable 3D human characters, enabling immersive virtual experiences and content creation. Metahumans are avatars that can be employed in a variety of applications, including video games, virtual reality experiences, film and television production. As lifelike digital avatars, Metahumans represent a significant leap forward in the realm of user experiences. When applied to video games, they enable gamers to immerse themselves in virtual worlds unlike ever before, enhancing their sense of presence and personal connection within the game.5

In the field of film and television production, they offer a revolutionary tool for storytellers and content creators. The ability to rapidly generate highly detailed digital actors with realistic facial expressions and movements can streamline the production process, reduce costs, and offer original creative avenues. Furthermore, Metahumans can facilitate the design of characters and scenes that were once limited by practical constraints (such as the need for physical actors, elaborate costume and makeup effects, and expensive set construction), which ultimately broadens the scope of storytelling possibilities (Jacko 2012).

These types of hyperrealistic representations also carry a notable downside, however. Deepfake videos are a type of synthetic media using artificial intelligence and machine learning techniques to create highly realistic but entirely fake audiovisual content. One of the most pressing concerns regarding deepfake videos is their potential for malicious exploitation. They can convincingly mimic the appearance and voice of real individuals, and can be used to deceive, or manipulate people on an unprecedented scale.

They can be employed to create counterfeit statements and endorsements from public figures, leading to misinformation campaigns that sway public opinion or damage reputations. Moreover, deepfakes pose a substantial threat to personal privacy. Individuals can have their likeness used without consent, making them unwitting participants in fabricated scenarios or offensive content.

In his book “The Post-Truth Era: Dishonesty and Deception in Contemporary Life,” Ralph Keyes (2004) examines the prevalence of dishonesty and deception in various facets of modern society, shedding light on the erosion of truth and the consequences of living in a world where authenticity becomes hard to discern (Keyes 2004). Hyperrealistic AI content amplifies these concerns. The pervasive blurring of lines between fact and fiction not only exacerbates the issues of dishonesty and deception but also underscores the urgency of addressing the post-truth landscape, in particular the potential for misuse and manipulation of information on a grand scale.

In an age where misinformation spreads rapidly through digital channels, the ability to create highly convincing yet fake content threatens not only personal privacy but also the very foundations of trust in informational sources. As a result, individuals and communities must grapple with the ethical and societal implications of these technological advancements, seeking ways to navigate this new digital landscape while preserving the integrity of information and truth.

How can we deepen our understanding of such an intricate technological realm and its impact on the multifaceted spheres of human existence?

I believe that the answer lies in artistic-practice-as-research. In particular, I am developing a project along with an interdisciplinary research team which aims to deliver immersive interactive experiences as multimedia installations. These installations will be presented in museums and public spaces. Physical and digital elements, such as video projectors, sound, and 3D animations will offer the visitors the opportunity to explore and interact with hyperrealistic AI content.

In fall 2024 and spring 2025, we will approach these interactive installations as case-studies, analyzing the emotional and cognitive impact of AI Metahuman characters, DALL·E images and AI voices on individuals. We seek to trigger pleasurable or discomforting sensations such as sympathy, happiness, anger, fear, grief, etc. by exposing visitors to AI generated images or sounds, categorized as emotional responses. We are also interested in visitors’ cognitive responses, such as thoughts and mental patterns that occur while they are listening, seeing and experiencing the aforementioned content. We believe that museums and public spaces serve as ideal settings for the presentation of AI interactive installations, since they enable visitors to explore the nuances of such a technology firsthand. The installations will be introduced to the public, who be encouraged to participate in our research by documenting their reactions through videos and photos while engaging with AI-generated content, thereby increasing awareness of the installations’ objectives and scope.

One of the installations will focus on a video projection presented in a dimly lit space. Visitors will be greeted by a colossal screen that seems to stretch infinitely in all directions. As the lights dim further, the projection begins, showing a state-of-the-art hyperrealistic AI coming to life with astonishing detail. A second installation will invite visitors to try an Oculus Rift headset for an immersive 3D animation experience that defies the boundaries of space and time. A third installation will focus on AI-generated voices that will activate a multi-dimensional soundscape conveying a spectrum of emotions, tones, and personalities, narrating dramatic stories for the public. Through strategically placed videos and speakers, these installations are intended to create an uncanny sensation of presence and proximity.

The installations aim to elicit a wide spectrum of emotions, from awe and wonder to introspection and contemplation. Visitors will find themselves transported to alternate realities, where they can witness the AI’s astonishing ability to replicate the natural world or to craft narratives that resonate deeply with human experiences. We will track the impact of these installations on the public through direct on-site observations and, whenever possible, through audio-visual documentations of the visitors’ reactions. Having collected data through the above-mentioned approaches, we will analyze the interactions between the public and the AIs through a sensorial and phenomenological examination of the participants’ subjective experience, conducting psychological evaluations (emotions, cognitive responses, understanding, interpersonal factors, etc.) and exploring neurophysiological responses (bodily movements, facial expression, skin/heartrate response and brain activity via mobile fNIRS).

The research outcomes will be presented through several outreach activities (exhibitions, public talks, workshops, publications) intended to reach multiple audiences, in order to make AI-related knowledge more accessible, breaking free from conventional limits which restrict information to AI experts, because of technical jargon or complex algorithms, data preprocessing, and computational resources that are typically beyond the reach of the general public.

As we navigate the uncharted territory of synthetic companionship and virtual experiences, it becomes clear that our relationship with reality itself is undergoing a profound transformation. This transformation, however, is accompanied by ethical and societal challenges. The societal impacts of hyperrealistic AI interactions are multifaceted. On the one hand, these technologies hold the promise to produce images so convincing that they can be harnessed for positive purposes like creating lifelike educational simulations, historical reconstructions or entertaining audio-visual content. On the other hand, they risk being exploited, for instance, in deepfake videos used for misinformation and manipulation, forgeries, identity theft and copyright infringements.

When approaching hyperrealistic AI content, we must strike a balance between benefiting from the advantages it offers and safeguarding against its potential hazards. We can embrace the innovation and creativity such technology offers, but will also need to implement comprehensive strategies and ethical frameworks to avoid negative impacts for our information landscape and digital trust. As we move towards a digital future that will increasingly be shaped by AI, it becomes necessary to comprehend the profound impact of this technology on both individual and societal levels. In this regard, the fusion of artistic practice and research not only enriches our understanding but also empowers us to make informed decisions, ensuring that we remain active participants rather than passive bystanders in the grand narrative of technological progress.

Jean Baudrillard, Selected Writings, Mark Poster (ed.), Stanford: Stanford University Press 1988.

Jean Baudrillard, Simulacres et simulation, Paris: Galilée 1981.

Rodney Brooks, “Robots at work: towards a smarter factory”, The Futurist 47 (3)/2013, pp. 24-27.

Rodney Brooks, “The Relationship Between Matter and Life”, Nature 409/2001, pp. 409-411.

Allyn, Bobby, “Surreal or too real? Breathtaking AI tool DALL-E takes its images to a bigger stage”, URL: https://wamu.org/story/22/07/20/surreal-or-too-real-breathtaking-ai-tool-dall-e-takes-its-images-to-a-bigger-stage/, accessed on July 20 2022.

David Chalmers, The Character of Consciousness, Oxford: Oxford University Press 2010.

David Chalmers, Constructing the World, Oxford: Oxford University Press 2012.

Ron Chrisley, “Embodied artificial intelligence”, Artificial Intelligence 149/2003, pp. 131–150.

Stanislas Dehaene, Consciousness and the Brain: Deciphering How the Brain Codes Our Thoughts, London: Penguin Books 2014.

Irina Ivanova, “Artists sue AI company for billions, alleging ‘parasite’ app used their work for free”, in CBSnews, January 20, 2023, URL: https://www.cbsnews.com/news/ai-stable-diffusion-stability-ai-lawsuit-artists-sue-image-generators/, accessed on July 7 2023.

Julie A. Jacko, The Human–Computer Interaction Handbook: Fundamentals, Evolving Technologies and Emerging Applications, Boca Raton: CRC Press 2012.

Ralph Keyes, The Post-Truth Era: Dishonesty and Deception in Contemporary Life, New York: St. Martin’s Press 2004.

Nora N. Khan, Seeing, Naming, Knowing, New York: The Brooklyn Rail 2019.

Lev Manovich, The Language of New Media, Cambridge: MIT Press 2001.

Giulio Tononi, PHI: A Voyage from the Brain to the Soul, New York: Pantheon Books 2012.

Fig. nr. 1: DALL·E generated images from the prompt: “Hyperrealistic portrait photographs.” Created by Pamela Breda, powered by GPT-4.

Fig. nr. 2: DALL·E generated image from the prompt: “Photograph of a mountain landscape.” Created by Pamela Breda, powered by GPT-4.

Fig. nr. 3: DALL·E generated image from the prompt: “Photograph of a classical sculpture’s head”. Created by Pamela Breda, powered by GPT-4.

© 2024. This work by Pamela Breda is licensed under CC BY-NC-ND 4.0. 

reposition ISSN: 2960-4354 (Print) 2960-4362 (Online), ISBN: 978-3-9505090-8-3, doi.org/10.22501/repos