During 2022 and the years preceding, generative Artificial Intelligence (AI) has made significant strides in experimentation and development. Generative AI refers to the type of AI that generates images and texts by responding to a specific text prompt entered by a human user. This technology is entering a new domain of human activity, and is expected to be the force that causes many notable changes. These changes are projected to affect the work of creative professionals, artists, and knowledge workers in particular. 

OpenAI, an AI research and deployment company that is a pioneer in the field of Generative AI, has had breakthrough developments such as the visual art generator DALL-E. In November 2022, the company released an open-source chatbot program called ChatGPT. ChatGPT can elucidate concepts in simple terms, write five-paragraph essays based on a prompt, and write poems and songs in a designated style. OpenAI is facing both criticism and anticipation from the public, from artists and academics’ respective wariness of AI’s appropriation of original artwork and potential threat to academic integrity, to the promising innovation and efficiency of a chatbot that could replace the search engine. 

How a Generative AI Generates 

Generative models are trained and systemized to learn by example via input data in order to imitate human judgment. Whereas discriminative models, which learn to discriminate between different data instances, are used for identification, generative models take a more complex route to produce new data instances. 

Professor Lee Sujin (Artificial Intelligence, Sejong University) explained that the basic model of generative AI in creating images called Generative Adversarial Network (GAN) consists of two networks, the Generator and the Discriminator. The objective of the generative model (G) is to generate near-real fakes and the objective of the discriminant model (D) is to determine whether the sample is fake or real. The ultimate goal of GANs, including these two models, is to generate data close to the ‘distribution of real data.’ 

Art Transcends the Artwork 

Whether or not this discovery should be viewed as a welcome revolution or a frightening infringement of creative human work, is dependent on the view one may have towards what makes the endeavor of art valuable. Due to machine learning, replication isn’t necessarily at the crux of Generative AI, as it is unlike photography that directly replicates the image behind the lens. New media technologies revolutionized art in their own way, and AI technology will follow in its own. 

Art transcends immediate sensory gratification, existing within human contexts; an identical pictorial structure does not guarantee the same artistic value. In the increasingly industrializing world of the 20th century and the subsequent burgeoning of replications, German philosopher and cultural critic Walter Benjamin coined the term aura to reflect how the specific time, cultural and social milieu makes each artwork truly valuable and particular. Another contextual value of art lies in the concurrent art scene, such as avant-garde art that breaks art traditions in innovative ways and defines new eras via new mediums, techniques, or concepts like Marcel Duchamp’s Fountain (1917) that later inspired conceptual art of the 1960s. 

AI art generators do not have the capacity to transcend the artwork, but only comply with human directions that help them do so. This in turn empowers humanity in this newfound method to deliver and express. “Technology plays a very big role in art,” commented Professor Lee, adding that it helps humans go beyond the limits of their possibilities to feel and express. “Tools for expression, starting from the hand to the pencil and the brush, are now leading us to experience ranges of infinity with numbers 0 and 1.” Art styles, such as oil painting, digital art, or watercolor, and distinctive styles of artists can be rearranged and reproduced, but curation of elements of the final work that gives it intention, context and social reverberations is the role of the human artist. 

Professor Lee suggests that one impact of AI technology on art would be the emergence of a new movement, forming a new genre of art. Generative AI will not threaten visual art but rather, challenge and diversify its scene in new ways. 

Knowledge Transcends Information 

Current AI lacks the singular human existence and agency that fundamentally informs both the input and output of creative work. This applies to the core skills that ChatGPT or other text or information generators are capable of. The human capacity for knowledge generation is not archival; it can be impulsive whereas AI can only be logical, and it can be orderly whereas AI can only be nonsensical. 

When employed as an information algorithm, ChatGPT will greatly accelerate information search and collection, proving to be a helpful tool for knowledge workers or students whose work requires information gathering. But lines may be blurred between information and original thought. Verbal arguments or expressions are products of each person’s personal experiences, social status, and repertoire of acquired frameworks of understanding both academic and practical, continuously bringing idiosyncrasy and novelty to constitute the different patchworks of discourse. Through language, certain truths are unveiled, newly produced, and come to acquire a new political, social, or cultural force. Generative AI utilizes the human reservoir of discourse to produce useful answers, but it cannot be a truly pioneering force. 

But that does not mean that ChatGPT is just another overhyped Silicon Valley ambition. As Professor Lee stated that AI techniques will “lead a paradigm shift in arts,” they may also shift the paradigm of creativity and creation. Because knowledge as humans know it transcends information and formalities, language-generating AI will challenge those in education to reevaluate assigning and grading written reports and essays as a process that necessitates experience and connective learning. This may further challenge the boundaries of plagiarism, original thinking, and the human episteme. 

Generative AI strongly showcases that human art and knowledge are both contextually formed and interacted with, requiring human intention and agency. The art and knowledge of generative AI is understood backwards, leaving more questions than answers about the legacy and future of human work. There will be even further advances in this new field in 2023, and deliberation, not fear or trivialization, is the required attitude for the advances of generative AI.