Executive summary. In his new book, How Great Ideas Happen: The Hidden Steps Behind Breakthrough Success, George Newman argues that the creative process resembles archaeology: surveying the landscape of ideas for fertile topics, digging for possibilities, and sifting through discoveries to uncover valuable insights. Within this framework, he describes AI as a powerful excavator of ideas, capable of rapidly exploring creative spaces and generating countless possibilities. However, he emphasises that AI still requires human guidance to identify which directions are truly worth pursuing. Despite the pace of technological change, the author contends that genuine innovation remains the domain of those with vision, discernment, and the persistence to look beyond the most obvious answers.

Key takeaways

• The creative benefits of AI depend almost entirely on how we use it:  In the hands of someone who can guide it thoughtfully, AI becomes a powerful tool for expanding creative possibilities.
• If you don’t already know what you’re looking for, AI just gets you mediocre results, faster.
• AI needs human curation, such as the ability to recognise which sparks hold promise and which should be left by the wayside.

---

The history of creativity is filled with tales of long walks, hot showers, and fitful dreams—those moments when our minds wander and a breakthrough idea suddenly seems to appear. Today, we have a powerful new ally in the search for those exceptional sparks. Generative Artificial Intelligence is revolutionising how we explore creative possibilities, allowing us to rapidly generate and test ideas in ways that would have been impossible just a few years ago.

Take humour, for example. One of the creative tasks I use in my lab is to have people write jokes. Participants are shown cartoons and are asked to come up with funny captions, similar to the popular New Yorker cartoon contest. Recently, I wondered, can AI be funny?

I showed the same cartoon to both humans and GPT—an image of business people sitting at a conference table underwater. I had people and GPT come up with captions to fit the cartoon. Then I gave all those jokes to a different group of people who rated how funny they were, without knowing which were created by humans and which by AI.

Both people and AI produced some genuinely funny captions, but there were telling differences in how they approached the task. The AI consistently gravitated toward puns: “When they said the company was going under, I didn’t expect this.” Humans, while occasionally using puns, showed more variety: “I know this isn’t a good time, but I’m going to need December 12th off,” and “This meeting could have been an email.”

But here’s what was particularly interesting: it took over 1,000 people, each generating multiple jokes, to arrive at the best human-created jokes. GPT-4, by contrast, produced a set of highly-rated jokes in a matter of seconds. This highlights a crucial pattern seen across creative domains: AI excels at rapidly exploring defined creative spaces, while humans are necessary for providing guidance and direction.

A similar pattern has now been replicated across many different labs and many different types of tasks. In short, when researchers have examined whether AI enhances people’s creativity, they find that in the hands of a person with expertise and a clear, well-considered creative vision, AI can be an extremely powerful ally, leading to work that, in some cases, surpasses what people can achieve on their own. But, without that careful direction, AI tends to drift into well-worn (and rather boring) creative territory.

Consider a study that examined how AI has impacted digital artwork. In general, the researchers found that the emergence of AI has led to less interesting art, overall. However, a small percentage of artists experienced the exact opposite effect: their work became dramatically more interesting and distinctive by using AI. For these artists, instead of allowing AI to dictate the subject matter, they appeared to use it to research different ways of realising their unique creative vision.¹

Moreover, in what is perhaps the most naturalistic study of how experts use AI to enhance their creativity and productivity, researchers at MIT asked college-educated professionals to do what they do best—either with the aid of GPT or not. For example, marketers came up with press releases for a product, HR managers composed a company-wide email, and consultants produced a consulting brief. In the end, professionals who were allowed to use GPT completed the task 40% faster, and the quality of their output—as rated by an independent panel of participants—was rated as 20% better. In the hands of skilled professionals, AI was clearly an asset.²

Together, this research tells us that the creative benefits of AI depend almost entirely on how we use it. When employed without clear direction—choosing familiar prompts or generic approaches—AI tends to produce boring, homogeneous results. But in the hands of someone who can guide it thoughtfully, AI becomes a powerful tool for expanding creative possibilities.

In my new book, How Great Ideas Happen: The Hidden Steps Behind Breakthrough Success, I argue that we can think of the creative process as being similar to archaeology: surveying the landscape of ideas for fertile topics, digging for possibilities, and sifting to see what we’ve found. Throughout history, the story of creativity has not been one of sudden lightbulb moments that are reserved for a chosen few, but rather, it more consistently emerges from curiosity, persistence, and a well-grounded sense of how (and where) successful ideas emerge.

Therefore, creativity is more of a skill that can be coached and refined than an innate talent bestowed on us at birth. We discover great ideas through collaboration, feedback, and attending to the world around us. Creativity is about what we do with the information around us, not who we are.

If you don’t already know what you’re looking for, AI just gets you mediocre results, faster.

And within this framework, we can think of AI as a powerful tool—an excavator of ideas. It can rapidly explore creative spaces and generate countless possibilities on a theme, but it needs human guidance to know which directions are worth exploring. It needs human curation, such as the ability to recognise which sparks hold promise and which should be left by the wayside.

So, does AI make us more creative? If we approach AI with the mindset that the goal is to search widely and clear as much ground as possible, leaving no stone unturned, then the answer (for many) seems to be ‘yes’. That said, AI still requires human expertise and ingenuity. If you don’t already know what you’re looking for, AI just gets you mediocre results, faster. As quickly as the creative landscape is changing, true innovation still belongs to those with vision, taste, and the willingness to keep digging past the obvious answers.

---

Footnotes

¹ Zhou, E. and Lee, D. (2024). Generative artificial intelligence, human creativity, and art. PNAS Nexus 3(3) pp.1-8 DOI:10.1093/pnasnexus/pgae052

² Noy, S. and Zhang, W. (2023). Experimental evidence on the productivity effects of generative artificial intelligence. Science 381(6654) pp. 187-192.