,

Artificial Intelligence’s Involvement in the Human Creative Process

,

By: Mbali Holt

Many researchers and scholars agree that AI promotes significant opportunities and potential for human creativity. As AI continues integrating into our everyday processes, it has been a resourceful tool for augmentation rather than completely replacing the human process. As generative tools greatly aid production and personalization, it is critical to maintain emotional depth, ethical use, and cultural relevance, which can only be truly captured by humans (Rubin, 2024).

Creativity is the ability to create something out of nothing. It is a power that integrates agency, knowledge, imagination, and desire to develop something that has not existed. Technology cannot perform the same creative process as humans. Instead, it replicates and imitates based on the patterns it learns from the entered training data, resulting in a lack of emotional depth and the human experience. AI is an augmented tool that acts as a collaborative partner in the creative process (Carroll, 2024). When it comes to creativity, technology cannot replace it, as AI relies on human creativity and judgment to train its models. Human processes and ideation remain a centralized unit of the technology's development. The technology serves as an amplifier of human potential as depicted in Figure 1. AI may aid with efficiency as it automates tasks and introduces new possibilities for human creative expression. The technology may "assist in the mechanics of creativity, but the heart of creative expression remains fundamentally human” (Nosta, 2024).

Figure 1: Infographic of AI's influence on human functions and processes, David Rice, “The Cognitive Cost of Convenience”.

Source: David Rice, “The Cognitive Cost Of Convenience: AI Will Impact Our Brains”

Many studies evaluate AI's impact on human creativity. For this paper, two studies are assessed to understand the effects of creativity on a human's innate creative process. One analyzes the influence of AI on creative writing, while the other evaluates the curation of research topics with the aid of generative models. The MIT Technology Review published an article that discusses AI's involvement in creative writing. A visual representation of the experiment  is displayed in Figure 2. Two researchers studied how people utilize OpenAI's ChatGPT to write short stories. It was determined to be helpful but had its limitations. To adequately analyze this experiment, the researchers focused on two metrics to measure creativity: usefulness and novelty. Usefulness demonstrated that the generated responses may be of use for publishable work. Novelty reflects the originality and uniqueness of the story. A total of 293 participants were involved in this research. Each person was tasked with writing an eight-sentence story for a young adult audience focused on one of the three proposed topics. The topics included adventures in a jungle, different planets, and the open seas.

The research team randomly sorted all participants into three groups. The first group had to create a unique story solely based on their innate creativity, the second group was given the ability to generate a story idea through ChatGPT, and the final group was allotted up to five AI-generated ideas from ChatGPT. Of all participants who offered to utilize AI, 88.4% used the tool. Next, they reviewed their work before a group of 600 reviewers took a look at their writing. Researchers determined that writers with AI access demonstrated more creativity. Writers who exhibited more significant levels of creativity did not benefit from a creative boost from the tool, while those who were less creative did (Williams, 2024). Less creative writers saw an increase of 10 to 11% in creativity and 22 to 26% in enjoyable content (Doshi and Houser, 2024).

Researchers concluded the experiment with three findings: the access to AI tools equalizes story evaluations, thus removing any advantages or disadvantages to one's innate creativity, no evidence concludes AI's ability to push beyond the creativity levels of humans, and lastly, many participants support the use of AI in creative spaces with the implementation of ownership, credit, and use-disclosure (Doshi and Houser, 2024).

Figure 2: Visual representation of the creative writing experiment utilizing AI.

Source: Anil Doshi and Oliver Hauser, “Generative AI Enhances Individual Creativity but Reduces the Collective Diversity of Novel Content”

The second study involved AI and participants tasked with developing research ideas. Participants were researchers who study natural language processing (NLP). NLP is a sector within computer science and AI that utilizes machine learning to allow computers to interpret human language (Stryker and Holdsworth, 2024). Researchers built an idea generator through a large language model called Claude 3.5, developed by Anthropic. They tasked the technology with finding relevant AI papers corresponding to seven research topics. The model generated 4,000 research ideas based on the discovered scientific papers. The AI-generated ideas were then compared to the research ideas of 49 human scientists. The scientists were tasked with the same goal of ideating research ideas based on the same seven topics. The researchers acted as reviewers and evaluated the feasibility, novelty, excitement, and effectiveness of all ideas, human and AI-generated. The reviewers deemed the AI-generated ideas more exciting and unique than those created by humans. Although the ideas may have been more intriguing, they were not all original. Researchers discovered that 200 out of the 4,000 AI-generated ideas were genuinely notable. Therefore, the model's outputs gradually became less original as it generated ideas (Conroy, 2024).

Although the tool has been demonstrated to be quite resourceful, it does have bounds and its use raises concerns. As ChatGPT was able to support writers with generating ideas, the generated stories tended to be similar to one another. Due to the similarities in semantics and content, as the model pulls information from the data it learns from, the generated responses become less distinctive instead of creating new and unique ideas. Claude-3.5 proved to generate novel ideas within limits as the model gradually became less creative with its outputs.

These experiments spark concerns surrounding the ethical use of AI in the creative industry, over-reliance on the tool, and its ability to outperform. Many are concerned with the infringement of ethical standards revolving around AI in the creative space. Models are trained based on human contributions, thus emphasizing the possibility of ownership and copyright infringement claims based on the tool not being inherently creative. They are trained based on learning from a human's innate creativity (Rubin, 2024). As the tool can transform the landscape of the artistic industry and enhance human creativity, there is the possibility of users solely relying on the tool for its potential outperformance of human-made work, as discussed in a webinar hosted by the Wharton School of Business (Basiouny, 2024). This reliance and themes of outperformance may hinder one's creativity,  development, and confidence. Users may also experience a "fixation of the mind" where once an AI idea is seen, it may be difficult for someone to think of one of their own (Habib, 2024). Therefore, evaluating the relationship between AI models, human originality, and creativity is essential. The tool may reduce the creation of innovative ideas and diminish the innate human imagination, thus stressing a balance within the relationship and encouraging its use as a tool rather than a replacement (Dekel, 2023).

In conclusion, as the world continues to navigate the intersection of AI and creativity, it is critical to balance the relationship between the two. AI has proven to be a resourceful technology that may aid in the creative process, serving as an augmentation tool. It may introduce novel ideas and opportunities to users. AI may support the brainstorming and ideation phases, enhancing human creativity. The tool plays a significant role but cannot yet completely replace, replicate, or interpret innate human creativity, emotional perception, cultural relevance, or problem-solving (Cai, 2024). These human characteristics remain central to the development of AI models. As challenges remain, such as reliance, ethical standards, and outperformance with AI, it is crucial to emphasize its use as an augmentation tool rather than a total replacement of the human creative process. It cannot wholly replicate the emotional depth, uniqueness, and creativity that stems from human imagination. Therefore, it solely serves as an innovation to amplify and expand human cognition and creativity.