The Question of Creativity in the Age of Artificial Intelligence
By: Rooz Aliabadi, Ph.D.
It’s becoming increasingly clear that Artificial Intelligence (AI) is evolving at an unparalleled speed, progressively demonstrating aptitude in areas initially considered exclusive realms of human intelligence, such as creativity. The recent study by Mika Koivisto & Simone Grassini explains this intriguing development, indicating AI chatbots surpass humans in average scores in the Alternate Uses Task — a recognized measure of creative thinking.
This groundbreaking study amplifies the ongoing, profound dialogue within the scholarly AI community concerning the implications and interpretations of computer systems excelling in endeavors initially and specifically designed to assess human capabilities. It brings forth many questions and reflections about what it fundamentally means for a machine to showcase aptitude in tasks rooted in human cognition and creativity.
These findings don’t necessarily display that AI is on the verge of acquiring intrinsic human abilities. AI’s capability to excel in creativity tests could merely reflect their programming and learning rather than an accurate indicator of inherent creativity as we humans understand and experience it. However, such pioneering research is invaluable, paving the way for a deeper comprehension of how humans and AI navigate and approach tasks demanding creative thinking and innovation.
The intricate methodology of the study involved examining three renowned AI chatbots — ChatGPT and GPT-4, developed by OpenAI, and Copy.Ai, built on the GPT-3 framework. These chatbots were instructed to brainstorm and conceptualize diverse uses for everyday items such as a rope, a box, a pencil, and a candle within confined time limits. The emphasis here was distinctly placed on the quality, originality, and creativity of the ideas generated.
Reflecting on this study, exploring AI’s potential was not limited to mere quantitative aspects but delved deep into the qualitative essence of creativity, with each chatbot undergoing multiple evaluations. Correspondingly, human participants were engaged under identical instructions, fostering a comparative environment to assess and understand the creative dimensions of both entities genuinely.
The evaluative approach was intricate, involving both algorithmic assessment and human evaluation, blending objectivity with subjective interpretation of creative nuances. Interestingly, while the AI responses received higher average ratings, the pinnacle of human responses reached higher creative zeniths, illustrating the nuanced capabilities inherent in human creativity.
Despite the enlightening revelations, the essence of this study was not to assert the supremacy of AI in creative domains or to propose their substitution for humans in roles demanding creativity. Instead, it serves as a philosophical exploration and a reflective mirror, prompting us to ponder deeply about the uniquely human qualities and the essence of being creative.
Simone Grassini’s reflection on the considerable technological strides in replicating human behavior over recent years resonates with me. The continuous evolution of these models highlights the potential intersections between human and artificial creativity. However, other perspectives that excelling in tasks designed for humans doesn’t confirm the machine’s capability of original thought also align with my viewpoint.
The tested chatbots’ inherent “black box” nature — our lack of complete understanding of their training data and response generation mechanisms — leads me to conclude that the models might not have genuinely created new ideas. Instead, they were potentially reiterating concepts seen during their training, making us question whether we are measuring true creativity or just recalling past knowledge.
Nevertheless, I find tremendous value in contrasting and comparing how machines and humans address problems. At MIT, Anna Ivanova’s insights are a pertinent reminder that minor alterations in the presentation of a task can significantly alter the chatbots’ performance. Such insights underscore the necessity of caution and reflection when interpreting the proficiency of AI in jobs rooted in human cognition.
In closing, these illuminating studies beckon us to investigate the intricate links between the tasks we assign to AI models and the cognitive faculties we endeavor to assess. We should approach this with humility and openness, embracing the diversity in problem-solving approaches between humans and AI models and avoiding the pitfalls of presuming uniformity in problem-solving methodologies.
In essence, this comparative exploration between AI and human creativity serves as a gateway to profound philosophical questions and reflections about the nature of creativity, the essence of human thought, and the potential and limitations of AI. It invites us to a dialogue, a mutual exploration, urging us to delve deeper into the essence of creativity and critically and reflectively explore the intersections and divergences between artificial and human intelligence.
This article was written by Rooz Aliabadi, Ph.D. (rooz@readyai.org). Rooz is the CEO (Chief Troublemaker) at ReadyAI.org
To learn more about ReadyAI, visit www.readyai.org or email us at info@readyai.org.
