Hu Yong: AI can generate everything, so do we still need human creators?

Question

**Author: Hu Yong, **Professor, School of Journalism and Communication, Peking University

Image source: Generated by Unbounded AI

The main use of ChatGPT is open dialogue, but people quickly found creative ways to use it, such as:

Answer Stack Overflow (programmer question and answer platform) questions
instead of google
Generate cooking recipes
Solve complex programming tasks
Generate image tooltips for Dall-e/Stable Diffusion
Build apps and websites from scratch

It seems that ChatGPT is very creative, however, the point is that ChatGPT is not specifically optimized for these uses, nor does it take into account its generality. Even so, the results on some specific tasks have been quite remarkable, giving many a glimpse of what might be on the horizon. That said, while these use cases won't scale ChatGPT into a general artificial intelligence, they could be useful in specific domains or applications very quickly.

There is no doubt that a new generation of artificial intelligence tools is taking the world by storm, tools that can help you write better, code faster, and generate unique images at scale. The advent of such powerful AI tools begs the question: What does it mean to be a creator in the age of AI ideas?

**I am inclined to think that if software like ChatGPT lives up to its grand promise, it can redefine human cognition and creativity. **

Generative Artificial Intelligence Challenges to Creativity

**AI creativity, which may also be called computational creativity, is a multidisciplinary field of research aimed at designing programs capable of human-level creativity. **

This field is not new. As early as the 19th century, scientists debated whether artificial creativity was possible. Ada Lovelace posed perhaps the most famous objection to machine intelligence: If computers can only do what they are programmed to do, how can their behavior be defined as creative? In her view, independent learning is an essential feature of creativity.

But recent advances in unsupervised machine learning do raise the question: Is the creativity shown by some AI software still the result of simply following instructions from human engineers? If Ada witnesses what the AI is already capable of creating, it may be hard not to doubt her original thoughts. As large language models (LLMs) got bigger, they started to deliver human-level, then superhuman results.

This has given rise to two schools of thought about AI creativity. **The first school sees artificial intelligence as a way to enhance human creativity—a creative partner for humans that inspires ideas, generates ideas, and overcomes creative blocks. The second group dreams of artificial intelligence that can imitate human creativity and become an independent creative thinker, capable of completely self-manufacturing and generating novel creative work. **

Some argue that since chatbots only learn statistical associations between words in their training set, rather than understanding their meaning, LLMs (Large Language Models) can only ever recall and synthesize what people have already done, and cannot exhibit some of the human aspects of the scientific process, such as creative and conceptual thinking. But won't it be like this forever? Will future artificial intelligence tools be able to grasp aspects of scientific processes that seem out of reach today?

In a seminal 1991 paper, researchers wrote that an "intelligent partnership" between humans and intelligent technology could transcend the intellectual capabilities of humans alone. These smart partnerships can accelerate innovation to previously unimaginable levels. The question is, where is the line between creative enhancement and creative production? How far can and should AI go when it comes to creativity? If AI can produce high-quality creations, is there a need for human creators?

AI cannot replace human creativity

As amazing as it is, **I think it is unlikely that AI will completely replace human creativity. First, creativity is a uniquely human trait deeply rooted in our biology and psychology. **It is the result of complex and unknown cognitive processes, such as pattern recognition, association, and synthesis, which cannot be easily replicated by machines. While AI can certainly be creative in some ways, it is unlikely to fully match the depth and breadth of human creativity.

**Secondly, at the heart of creativity lies the ability to express emotions and experiences in unique and personal ways. **AI algorithms cannot truly understand the human experiences and emotions that fuel creativity. An AI writer will be able to piece together enough human experience to tell a convincing story, but there is something inherently human about the reader’s experience of the creator in knowing that the author experienced the pain and pleasure they describe, that a machine writer cannot. The personal touch that human creators bring to their work adds a level of authenticity that AI algorithms simply cannot match.

Many artificial intelligence researchers are debating whether to create machines with emotions. Emotions are absent in typical AI models, and some researchers say programming them into machines could give them a mind of their own. Emotion, however, sums up human experience because it enables humans to remember those experiences. "No computer can be creative unless it can simulate all the nuances of human emotion," writes Yale computer scientist David Gelernter.

**Again, creativity often involves capturing human sensitivities such as context, perspective, and cultural nuances. **While AI algorithms can be trained to recognize patterns and generate content based on data, they cannot understand human sensitivities in the same way that humans do. This also makes it difficult for AI to deal with topics on which society does not have a general consensus, such as political and religious issues. If you try to generate text on these topics, you may end up with biased, inaccurate, or outdated text.

**The fourth important reason why artificial intelligence cannot replace human creativity is lack of imagination and intuition. **Creativity requires the ability to think beyond existing things and imagine new possibilities, which AI algorithms lack. They can't come up with truly unique and original concepts. Just because AI isn't really the creator, one of the big downsides of the content it generates is that it's not entirely original. Content generators simply gather information that already exists within certain parameters. Therefore, while the content will pass plagiarism checks, it will not include original research, insights or data. In practice, this means it has no ability to share ideas or create thoughtful content.

**Finally, the unpredictability of creativity is another factor that sets it apart from AI. **Creativity can be unpredictable and spontaneous, involving sudden flashes of inspiration, experimentation, and serendipitous events. However, AI algorithms are limited by the programs and data they are trained on, and lack the ability to react in real time to new information, so it is impossible for them to fully replicate the unpredictability of human creativity.

All in all, generative AI cannot replace human intelligence and insight. To be truly original, generative AI needs to be guided and nurtured by human creators with domain expertise and background experience. By providing the right hints, human creators can help generative AI reach its full potential, producing impressive results. So while generative AI is a very powerful tool, it is still just a tool that relies on human creativity, expertise, and experience to be truly effective.

Strengths and weaknesses of artificial intelligence tools

While we acknowledge the instrumental nature of AI, the problem is that most people have little AI literacy—an understanding of when and how to use AI tools effectively. What we need is a straightforward, common framework for evaluating the strengths and weaknesses of AI tools, accessible to everyone. Only then can the public make informed decisions about incorporating these tools into our daily lives.

To meet this need, we might as well use an old method in the field of education: Bloom's taxonomy. This taxonomy was first published in 1956 by educational psychologist Benjamin Bloom and was later revised in 2001. It is a hierarchy describing levels of thinking, where higher levels represent more complex thinking. Its six levels are:

Intellectual memory: acknowledging or remembering facts, terms, basic concepts, or answers without understanding their meaning.
Comprehension: Explain main ideas and concepts and express meaning by explaining, classifying, summarizing, inferring, comparing and clarifying.
Application: Use knowledge to solve problems, identify how things are connected and how they apply in new situations.
Analysis: Examining information and breaking it down into its component parts, determining relationships between parts, identifying motives or causes, making inferences, and finding evidence to support generalizations.
Evaluation: Making and defending opinions based on a judgment about the validity of information, ideas, or quality of work based on a set of criteria.
Creation: Putting elements together to form a coherent or fully functional whole. This is the highest level of Bloom's taxonomy.

Bloom's taxonomy is not tied to a specific technology - it applies broadly to the cognitive domain. We can use it to assess the strengths and limitations of ChatGPT or other AI tools that manipulate images, create audio, or fly drones.

In general, ChatGPT does well on memory, comprehension, and application tasks, but struggles with more complex analysis, evaluation, and creation tasks. For example, if we use Bloom's taxonomy to observe the professional future of doctors, lawyers and consultants, we will find that artificial intelligence may one day reshape these professions, but not completely replace them. While AI may be good at memory and comprehension tasks, few people ask a doctor for all the possible symptoms of their illness, ask a lawyer to explain the letter of a law word for word, or hire a consultant to explain Michael Porter's five forces.

In those higher-level cognitive tasks, we turn to experts. We value the clinical judgment of physicians in weighing the benefits and risks of treatment options, the ability of attorneys to synthesize precedent and mount a vigorous defense on our behalf, and the ability of consultants to identify out-of-the-box solutions that no one else has thought of. These skills pertain to the tasks of analysis, evaluation, and creation, a level of cognition that is currently beyond the reach of artificial intelligence technology.

Using Bloom's taxonomy, we can see that effective human-AI collaboration will largely mean delegating lower-level cognitive tasks so that we can focus our efforts on more complex cognitive tasks. So rather than dwelling on whether AI can compete with human creators, ask how AI capabilities could be used to help develop human critical thinking, judgment, and creativity.

Of course, Bloom's taxonomy has its limitations. Many complex tasks involve multiple levels of taxonomy, frustrating attempts at classification. And Bloom's taxonomy doesn't directly address bias or hate, a major problem in large-scale AI applications.

But, while not perfect, Bloom's taxonomy is useful. It's simple enough that everyone can grasp it; general enough to apply to a wide range of AI tools; and structured enough to ensure that we ask a consistent set of thorough questions about those tools.

Just as the rise of social media and fake news requires us to develop better media literacy, tools like ChatGPT require us to develop our AI literacy. Bloom's taxonomy provides a way to think about what AI can and cannot do as this type of technology becomes embedded in more parts of our lives.

I choose, therefore I am

Interestingly, generative AI seems to create an urgent need for human creativity. It's easy for an AI to just randomly come up with something novel. **But it is very difficult to come up with something new, unexpected, and useful at the same time. **

However, the paradox is that, with generative artificial intelligence to rely on, human creativity may enter a trough. In July 2019, during a chess match in France, Igors Rausis, the 53rd-ranked international grandmaster in the world, was exposed to using a mobile phone during the match, which is considered cheating according to the rules. Garry Kasparov, the first chess world champion in human history to lose to a computer, commented that although using a mobile phone in real life is not cheating, you may develop cognitive deficits due to excessive reliance on digital crutches.

He emphasized that if we only rely on machines to tell us how to be good imitators, we will never be able to take the next step and become creative innovators. Similar to our bodies, our brains need exercise and are constantly trained by performing demanding and challenging cognitive tasks in order to excel and spark that "Aha!" insight.

Unfortunately, once we delegate some cognitive autonomy to intelligent machines, it will be extremely difficult to get it back. That’s why, while humans are putting the brakes on their enjoyment of cognitive journeys, algorithms and artificially intelligent machines are advancing at incredible speed, serving as new sources of creativity. Some have a utopian vision of the fully automated AI future we are entering at a rapid pace, while others have a hysterical vision of it. **In this case, each of us has a choice: to embrace these new challenges, or to contain them. Are we going to help shape the future, setting the terms of our relationship with algorithms and intelligent machines, or let algorithms and intelligent machines impose them on us? **

In his brilliant 1976 book, Computing Power and Human Reason, Joseph Weizenbaum argued that "no matter how intelligent a machine may be, there are certain acts of thought that can only be attempted by a human being." He extolled the importance of judgment, intelligence, and compassion—things that we cannot outsource to machines, even if we could. In a profound formulation, he wrote that machines can decide, but they don't choose. Why does a machine do what it does? Every mechanized decision can be traced back step by step through an algorithm, until it finally reaches an inevitable conclusion: "Because you told me." For humans, this is not the case, the fundamental explanation is: "Because I chose." In this simple phrase, there is human agency, human creativity, human responsibility, and human beings themselves.

We've argued that our technology can make us more human and free us to be more creative, but there's more to being human than creativity. We have other qualities that machines cannot match. They have instructions, and we have purpose. Machines cannot dream, not even in sleep mode. Humans can, and we will need our intelligent machines, in order to turn our greatest dreams into reality. As Kasparov said, if we stopped dreaming big, if we stopped looking for something bigger, then we might be machines ourselves.

Creativity has long been considered one of the main pillars of anthropocentrism. Apart from language, values, emotion and perception, what would make us human if not creativity?

View Original