Human Creativity vs AI

Generative AI is so impressive that it challenges the belief in human exceptionalism like nothing else before. Is there something that humans do that nobody or nothing else can do? Have we created a technology that is equal to or even better than us in every aspect of human activity? Is the human mind nothing more than a tool, and not even the best one?

In a brilliant and thought-provoking essay, Donald Clark argues that generative AI has shattered the illusion of human exceptionalism. Although on a fundamental philosophical level, I still adhere to my beliefs in human exceptionalism, it's hard to deny that the question of "what is human" (aside from "to err") has never been so tangible.

Optimistic articles abound on the web, claiming that generative AI is not here to replace human intelligence and steal our jobs, but to assist us in our work and enhance our abilities. These articles seem overly naive to me. This is a "boiling frog" talk. Elon Musk and other experts didn't write the letter urging a pause in AI development for amusement. They certainly know what they are talking about.

In its early teenage years (that is, now), AI is already shockingly proficient in knowledge, reasoning, and creativity. With minimal or no coding experience, using ChatGPT you can now develop your own Web application or a simple video game in a matter of hours. Sure, the process is currently a bit cumbersome, but considering the progress AI has made in just the last 7 months, it is unrealistic to hope that progress will stop here or slow down. The jokes that ChatGPT generates at the moment may not be funny, but they will improve. Tunes generated on Aiva.ai or similar tools may not win a Grammy award yet, but they are good enough for jingles or even mediocre soundtracks. Midjourney can produce digitally remarkable and artistically valuable art. Video-generating AI tools are in development.

So, it seems crucially important to determine what is left for humans (if anything) so that we can start preparing for the near future, especially us, the educators. Whatever is unique to humans and unattainable by AI is what education should focus on.

First, we must acknowledge that generative AI is creative. It could just as well be called Creative AI. Creativity is a capacity that involves generating new and valuable ideas, and AI possesses it.

One might argue that generative AI is merely a synthesizer of existing ideas, but so is the human mind. Nothing is truly new under the sun, and creative synthesis is the core process of human consciousness.

So, how is human creativity different from AI creativity? It's a philosophical question, and it would make sense to seek answers within the realm of philosophical thought.

Two relevant epistemological theories to this question are empiricism and rationalism. Empiricism claims that knowledge primarily comes from experience (the five senses), while rationalism emphasizes reason and logic in the creation of new knowledge.

Rationalism and empiricism are not mutually exclusive. New knowledge can be created through pure reason (or math), yet the primary concepts involved in reasoning must be derived from the experience of reality. All the ideas that pure reasoning can generate are syntheses and derivatives of primary concepts. For example, a Chinese sour-sweet sauce is a culinary innovation, but the concepts of "sweet" and "sour" are derived from sensory experience. Humanism, as a philosophy and system of values, is based on empathy, which is derived from emotions triggered by social interactions. The theory of general relativity is based on the observation of light bending around Mercury.

To have an edge over AI in creativity, humans must focus on the primary concepts because AI can reason but cannot experience. If ChatGPT was trained on the data accumulated up until 1542 (a year before Copernicus discovered the heliocentric system), it would still assume that the Sun is orbiting around the Earth, as the discovery of the heliocentric system required extensive observations and a bit of serendipity in addition to mathematical calculations.

Therefore, I believe that the empiricist view of knowledge creation provides at least a hint as to what kind of creativity is uniquely human. It is not a coincidence that among the notorious 5 skills of disruptive innovators by Dyer, Gregersen, and Christensen, 3 are "empiricist" (observing, experimenting, and networking), and 2 are "rationalist" (questioning and associating).

In education, creativity should be practiced and developed accordingly, with a focus on creating meaning from experience.

Learners should engage with the world to create new meanings derived from experiences rather than existing ideas. ChatGPT can outperform any individual in the Alternative Uses Test or Remote Associates Test. It can approach an existing problem from a thousand different perspectives, make all the associations, and identify all the patterns in existing knowledge.

What learners must focus on is going out "in the field" and bringing back new meanings gained through real experiences. There is a lot to discover out there, and there always will be because the world is complex, dynamic, and probabilistic. Therefore, there will always be a need for new valuable ideas based on hands-on experience.

When searching for novel experiences, follow the problems. Wherever there is an unresolved problem, old or new, there is an opportunity for to gain creative insight. According to the "Anna Karenina" principle, there are multiple ways in which things can go wrong, and there is only one way to get things right. Therefore, it is much more likely to gain novel experiences from problems rather than existing solutions, from things that don't work rather than those that work.

Search for authentic creative experiences. An important difference between AI and humans is that AI's actions are deterministic (an algorithm determines how inputs convert into outputs), while humans are capable of volitional acts and free choice. Additionally, any AI tool (e.g., ChatGPT) has only one Large Language Model behind it, while billions of humans each have a unique and large "model" in their brains. Thus, experience driven by authentic motives (passion, values, curiosity, empathy, etc.) is particularly human and is likely to generate creative insight that AI cannot produce.

Concretely, in terms of pedagogy, the ways to develop and practice human experiential creativity in educational context are to adopt challenge-based learning, service-learning, inquiry-based learning, and other methods that involve learning through real-life "field" experiences.

It is certainly not certain that one day AI won't acquire all the remaining abilities that are for now uniquely ours. But even then, I won't accept the idea of human non-exceptionalism. Because, at the end of the day, it is humans who created AI. How is this not exceptional?