In 1928, the Scottish scientist Alexander Fleming returned from a two-week vacation and realized he had made a significant oversight. In his haste to leave for his holiday, he had left a messy pile of petri dishes on his workstation. As he began to dispose of the dishes, something unusual caught his eye: a mold had formed around the bacteria in one of the petri dishes.

Upon closer examination, Fleming made a startling discovery. The mold wasn’t just growing—it was actively inhibiting the spread of the bacteria. This observation led Fleming to wonder if the mold could be used to fight infections. Over the following weeks, his experiments with this novel substance led to the development of penicillin, revolutionizing the treatment of infectious diseases, saving over 500 million lives over the past century.

Penicillin’s unlikely origin story holds a key insight into how human creativity may continue to flourish in the age of artificial intelligence. The past year has seen a surge of concern—some might say bordering on hype or hysteria—about AI’s potential to take over many jobs. In January, for instance, the International Monetary Fund reported that up to 40% of global jobs were at risk from AI. From cashiers to coders, workers across various sectors are being replaced by machines.

This raises a critical question: If the computational powers of generative AI continue to grow, where does the human brain maintain an enduring advantage? The answer may lie in the 1928 story of Fleming’s discovery: The human capacity to make and learn from mistakes is a fountain of innovation. “Do not be afraid of mistakes,” Fleming once said. “For without them, there can be no learning.”

Human fallibility, rather than being a liability, is an asset when it comes to creativity. It is our ability to embrace randomness, luck, and flexibility that often leads to breakthroughs. Fleming himself acknowledged this, saying, “One sometimes finds what one is not looking for. When I woke up just after dawn on Sept. 28, 1928, I certainly didn’t plan to revolutionize all medicine by discovering the world’s first antibiotic. But I guess that’s exactly what I did.”

While AI excels in efficiency and computational power, humans are slow and sometimes inefficient processors of information. However, what we lack in speed, we gain in freedom—the ability to disregard conventional wisdom in pursuit of originality.

Consider a field like investing. There is no doubt that AI will become vastly more capable than any single individual in digesting and interpreting vast sums of industry and company data. And perhaps AI may react more rationally than a human during times of market volatility.

But in investing—as well as many other disciplines—outperformance often relies on unconventional thinking and going against the grain. In fact, as AI tools make gathering consensus information and existing data more efficient, the role of human intuition could arguably become increasingly important to outperform over the long-term.

The same goes for another profession like reporting: While AI could make the journalistic research process more efficient, simply repackaging company press releases and compiling existing data won’t earn any journalistic accolades. Spending time in person will often be necessary to get to the deeper roots of a story. Consider how an adept journalist (or even investor) might detect a nervous tick or read certain visual cues from a cagey CEO, which inspires them to dig further into a story. Or perhaps how an in-person meeting with a source might make them feel comfortable enough to share sensitive information.

Earlier this year, I wrote an essay exploring the human and behavioral elements of both investing and journalism: “The best reporters and the best investors have a lot in common: a high threshold for pain, antisocial behavior, and a strong BS detector.” In other words, human intuition will always play a critical role in understanding the deeper context of a situation.

The importance of softer skills—like compassion, emotional intelligence, cross-cultural sensitivity, and strategic thinking—is increasing. At the very least, these skills will become ever rarer in a landscape dominated by AI-enabled systems. But AI-driven robots don’t yet have the capacity to interpret complex social dynamics.

In April, researchers at the University of Birmingham explored the nascent idea of “cultural robotics.” As the authors write, this idea “aims to design robots that can take into account the cultural background of the person they’re talking to, and adjust their behavior accordingly.”

The authors cite studies exploring cultural differences regarding personal space, facial expressions, and other uniquely human elements of interpersonal relationships. “Different cultures interpret facial expressions differently too,” the authors note. “One study found that people are more able to understand a robot if it communicates using the facial expressions that they are familiar with.”

The stakes get even higher in a field like healthcare, where AI could potentially present a disruptive technological shift to the status quo. “From the development of drugs and vaccines, to improving medical diagnosis and treatment, [AI is] being used in all stages of the value chain, boosting efficiencies across the overall healthcare system,” the World Economic Forum noted last June.

And yet, despite the vast efficiencies and positive medical outcomes from integrating AI into healthcare, some of the most sensitive parts of the healthcare system depend on human, in-person relationships. A nuanced understanding of emotions—i.e. emotional intelligence—enables experienced medical professionals to provide patients and families with comfort.

It’s also a necessary component to establish trust that’s so vital to the doctor-patient relationship.

Healthcare is just one field where “soft skills” are still vastly more capable than AI. But taking an even wider view, consider concepts like art and happiness.

On one hand, yes, an AI will be able to regurgitate vast quantities of information about the world’s most famous artists or filmmakers. It will know every recipe for every dish ever made. And it will have the computational ability to have listened to all music ever recorded.

But does it know what it feels like to tour the vast halls of the Louvre? Has AI ever experienced a homemade meal? Can it possibly know the joy and wonder of seeing a musician on stage perform a rock ballad before thousands of screaming fans?

This is a key point, considering that several studies have linked happiness to an increased ability to generate new ideas and increase creativity. Dr. Simone Ritter, who studies the relationship between the brain and human creativity, found in a 2017 study that listening to “happy” music can actually lead to more creative thinking. “Being in a positive mood facilitates flexible thinking and consequently leads to the production of unconventional or atypical ideas,” Ritter notes.

Back in 1928, when Alexander Fleming made his remarkable discovery, he did so as the result of a number of human-centric foibles. For those workers who are concerned about AI, the antidote is perhaps to lean into the traits that make us uniquely human. Continue to travel, continue to be curious about the world, continue to learn and fail, and continue to embrace the “soft skills” centered around developing emotional intelligence. The potential for profound human innovations—from the Italian Renaissance to the invention of the internet—were fueled by a blend of cultures and uniquely human experiences.

This should give you an edge—for now, at least.