Serendipity in Science: What’s its Fate in the Age of AI?

They call it by different names: happy accidents, unexpected discoveries, lucky breakthroughs—Serendipity. And I’ve experienced it firsthand. No, I haven’t stumbled upon penicillin or graphene, but I did discover a DFT functional that could mimic dispersion forces in systems of biological relevance. Back then, it was a big deal. I’d wager that every scientist, whether tinkering in labs or poring over simulations, from DNA sequences to polymer chains, has brushed up against serendipity. You dive deep into one question, obsessed with untangling your scientific yarn, only to come across an entirely different knot. Think of the roots of string theory. It set out to explain the strong nuclear force tying quarks together within protons and neutrons. Yet, gravity kept weaving itself into the equations, turning string theory into a potential theory of everything. Mind-blowing, right? Now, with all my writing about quantum chemistry, simulations, and AI-driven discovery, I’ve found myself deep in thought about the nature of serendipity. So, I set out to craft a story that might shed some light, offering up answers both to you and to myself. Instead, I’ve landed in a place teeming with more doubts than I began with. So, what’s serendipity? Where does AI fit into the picture? And what can we anticipate for the future of scientific discovery? Let’s go.

Serendipity in Science: What's its Fate in the Age of AI? | From Atoms To Words | Arturo Robertazzi
Emerging serendipity

Serendipity: A Story of Scientific Discoveries

History is filled with accounts of radical ideas and innovations that didn’t exactly play by the rules. These were ideas born not just from hitting the books hard but also from a mix of curiosity, happy accidents, and a whole lot of try and try again. Think about it: the accidental discovery of x-ray emissions from peeling tape in a vacuum, the famous random walk to graphene, or the realization of cisplatin‘s potential as an anticancer agent—each reveals how unexpected findings have revolutionized scientific discovery time and again.

Take Alexander Fleming and his not-so-tidy lab that brought us penicillin, or Albert Hofmann who got more than he bargained for with LSD. Instances like these underscore the importance of keeping an open mind and embracing the What’s this? moments in science.

The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! but Uh, that’s funny…

Isaac Asimov

Serendipity itself is quite the story. Back in 1754, Horace Walpole stumbled upon an old Persian fairy tale called The Three Princes of Serendip. As these princes wandered, they kept bumping into incredible discoveries. They weren’t looking for it; it just happened. Walpole thought this was the perfect way to describe those aha moments when you find something amazing you weren’t even looking for.

And that’s serendipity for you: tripping over a breakthrough while heading somewhere else entirely.

Now, serendipity has been (not so quietly) making its mark across the board, nudging history, science, and even our personal lives in directions we never saw coming. Which is kind of magical.

So, here we are, pondering—is serendipity just a fancy way of acknowledging our luck, or does it serve as a crucial spark in the engine of scientific discovery? Is serendipity a feature of human creativity or a measure of our ignorance?

More on From Atoms To Words:
Digital Alchemy: Computers in Chemistry and the Future of Scientific Discovery
Chemical Space to Material Discovery: Simulations and Machine Learning Leading the Way
Do We Live in a Simulation? Yes, No, Maybe: Insights from 12 Leading Experts

Serendipity: Ignorance or Human Trait?

Ever heard of super-encounters? These are folks with a natural knack for serendipity, those who eagerly dive into oceans of data, hopeful to unearth hidden gems. These individuals possess an uncanny ability to stumble upon unforeseen insights, even while leisurely flipping through the volumes of a library or meandering among shelves of organized knowledge. As we navigate this AI age, how crucial is it to design digital environments that foster rather than stifle this spontaneous discovery? And is finding the right balance between data-driven research and openness to the unexpected the key to fueling innovation? Let’s explore these questions together.

▸ Serendipity as a Measure of Human Ignorance

Remember our story about the infinity of chemical space? Just for small organic molecules alone, all possible combinations stretch from 1022 to 1060. And when you throw in the mix of organic and inorganic compounds—forming films, fibers, membranes, gels, or complex superstructures—the scale of permutations dwarfs the number of atoms in the universe. Quite literally.

What if we fed every tiny bit of the chemical space into a super-smart AI model? Would we conjure up the ghost of a Laplace’s AI-Demon? The thought is terrifying, but don’t worry—it’s just a thought experiment. Mapping the entirety of chemical space is an unimaginable task, skirting the bounds of the impossible.

Now, humor me for a moment.

If we somehow edged closer to a full knowledge of the chemical space, would serendipity still find a way into the heart of scientific discovery?

The rise of machine learning and AI, fed by a sea of big data, hints at a future where serendipity’s flame might dim. We could be inching toward a world where everything is known, leaving no room for chance. Welcome to the Laplace’s AI-Demon’s era.

Chance favours only the prepared mind

Louis Pasteur

But then again, consider the sheer amount of data we’re sitting on today. Far from reaching the Laplace’s AI-Demon’s capabilities, think of GNoME, an AI model that has uncovered over two million new crystal structures. Does this treasure trove aid materials science?

Yes and no.

Despite this wealth of information, extracting meaningful insights is like finding a needle in a haystack. The needle might be slightly bigger, or the haystack a tad smaller, yet the challenge persists. Sure, you have 2 million new structures, but are they feasible? Useful? What properties do they possess? What I’m saying is, even with all this data at our fingertips, we’re still in the dark on many fronts. We are merely shifting our ignorance from one place to another.

Even as AI—be it AlphaFold, large language models for chemistry, or machine learning tools for materials science—presents us with answers on a silver platter, we remain largely ignorant about the underlying process.

True science is as much about the journey as it is about the destination. Having the right results is handy, but without the journey, our knowledge feels incomplete.

Could it be, then, that even beneath today’s towering mountains of data, the serendipitous spark of discovery might still find its way, illuminating unexpected paths in the dark unknowns?

▸ Serendipity as a Feature of Human Creativity

On the flip side, there’s something truly magical about the concept of serendipity as a feature of human creativity. Design research confirms that groundbreaking innovations often emerge not from meticulous analysis but from a chaotic fusion of ideas, driven by curiosity and a quest for knowledge. This synergy between creativity and serendipity distinguishes real innovation from mere iteration.

I cannot help but think about the random walk to the discovery of graphene by Geim and Novoselov. Yes, in essence, it was a happy accident. This does not imply that it wasn’t a product of deliberate scientific effort; rather, it highlights that Geim was not actively seeking it—at least, not initially. Geim organized these Friday Night Experiments, a playful tradition aimed at exploring uncharted territories away from his group’s main research focus. And from this openness to exploration, significant discoveries emerged. There were the magnetic properties of water and the infamous floating frog, leading to an Ig Nobel Prize. More notably, there was the Scotch Tape episode that led to the discovery of graphene.

Geim’s Friday Night Experiments created an environment where serendipity could thrive, almost as if summoning the gods of science. Geim and Novoselov’s genius lay in their ability to recognize and embrace these unexpected moments of discovery, proactively nurturing serendipity.

When one dares to try, rewards are not guaranteed, but at least it is an adventure

Andre Geim

The graphene story exemplifies how the interplay between serendipity and creativity thrives on diversity and openness. Exposure to different stimuli and perspectives boosts the odds of scientific happy accidents, suggesting that humans indeed flourish in varied, unstructured environments that promote creativity and, by extension, serendipity.

If you think about it, serendipity might just be the essence of how we, as humans, approach science. Perhaps it’s less about the volume of data at our disposal and more about how we perceive the world through our inherently human senses.

Could this be the reason serendipity plays such a pivotal role in scientific discovery? If so, then shouldn’t our goal be to foster serendipity and create AI environments where it can thrive?

Serendipity in Science: What's its Fate in the Age of AI? | From Atoms To Words | Arturo Robertazzi
Serendipitous patterns in an Au-Cu alloy model | QuantistryLab

More on From Atoms To Words:
When Will RNA Structure Prediction Get Its AlphaFold Breakthrough?
Machine Learning in Materials Science: A Second Computational Revolution?
Is Machine Learning Going to Replace Computational Chemists?

So, What’s the Fate of Serendipity in the Age of AI?

It’s no secret that machine learning and AI have been game-changers in the sciences: from protein structure prediction to drug discovery and material research, this is just the beginning. But as we admire this leap forward, let’s pause and wonder: What becomes of serendipity? Of those unexpected discoveries that have historically sparked radical innovation? Here’s the crux of the matter: Does our drive for data efficiency and precision risk sidelining the serendipitous explorations that lead to groundbreaking findings?

AI shines in environments with clear rules, like games, where outcomes are cut and dry. Remember AlphaGo’s victory in 2016? A blast for sure, but the first notable triumph of AI over a human traces back to 1997, with Deep Blue’s victory over chess champion Garry Kasparov.

But when it comes to the wild prairies of innovation, where breakthroughs often pop up from the least expected places, sticking to the script isn’t enough.

Now, think about protein structure prediction and the success of AlphaFold. This AI model crunches through mountains of structural data to deliver accurate insights. In contrast, as we’ve seen in our recent story, RNA structure prediction may be decades away from being solved. The main reason? The lack of structural data for RNA compared to proteins. The bottom line? AI in scientific discovery is its own unique story.

This brings us to a pivotal point: perhaps the true potential of AI isn’t about brute-force data crunch to replace human ingenuity; it’s about amplifying it.

In the age of AI, serendipity in science will flourish, not fade

ChatGPT4

Consider, for example, the knowledge feedback loop, a cyclical process that iteratively enhances understanding and insights by generating, sharing, applying, and refining knowledge based on continuous feedback. When applied to scientific discovery, we can envision AI as a kind of scientific sidekick, formulating hypotheses, assisting in experimental or computational design, and sifting through results not only to provide a quick answer but, perhaps, most importantly, to ignite new ideas.

And then there’s the frontier of computational creativity. Yes, that is a thing. Computational creativity is used to model imagination in fields such as the arts and music. The same approach could be applied to science to diversify research directions and help humans ponder over novel paths through novelty-oriented algorithms.

This gets us thinking, doesn’t it?

The blend of dry AI with the magic of serendipity may be ushering in a new chapter where intentionally designed information systems do more than just meet our needs—they could fling open the doors to the unknown unknowns.

Despite AI’s structured approach, perhaps the spirit of serendipity is not dead after all. Who knows? In the age of machine learning, serendipity might just be getting started.

Serendipity in Science: What's its Fate in the Age of AI? | From Atoms To Words | Arturo Robertazzi
Serendipitous patterns in a protein

More on From Atoms To Words:
AI in Drug Discovery: Chasing Dreams, Facing Realities
Large Language Models for Chemistry: Is the Beginning of a New Era?
60 Years in the Making: AlphaFold’s Historical Breakthrough in Protein Structure Prediction

A Final Personal Touch

So here we are, at the end of a journey that’s taken us a bit off the beaten path from our usual down-to-earth topics. My aim with this story was to unpack how AI might just outpace serendipity. You know, with all this big data at our fingertips, I wondered if those eureka moments in scientific discovery might start to fade into the digital background. And I’ve got to say, part of me still buys into that—believing that a more in-depth analysis and intelligent interpretation of data could lead us to more deliberate discoveries, reducing our reliance on chance.

As I kept writing, my confidence in that idea began to shake.

Picture this: the wider our base of knowledge, the richer the playground for our scientific curiosity to roam free. Within that context, serendipity itself might even grow stronger. Scientific discovery, even in the age of big data, may still be at the mercy of chance, randomness, and yes, serendipity.

Maybe, just maybe, in this brave new AI world, serendipity isn’t facing its end but is on the brink of a renaissance, ready to unlock doors to discovery, learning, and innovation in ways we’re only beginning to understand.

Who knows, perhaps serendipity will evolve into some sort of metaserendipity—a form of serendipity enhanced by today’s endless and ever-growing rivers of data. This funky concept (the result of a late-night caffeine-induced inspiration) suggests the importance of intentionally crafting digital spaces that seek not only answers but also invite the unexpected. A new era, where technology and design join forces to lay the groundwork for serendipity to flourish, especially when we’re swamped with more information than ever. By encouraging exploration that goes beyond the linear, metaserendipity could push us towards a systematic integration of curiosity and chance into AI-driven knowledge.

After poring over my thoughts across these pages, I’ve come to see that rather than coming to a neat conclusion, I’m walking away with more questions than answers. It’s late now. Rain taps against the window, a rhythmic reminder of the world spinning forward.

As I sit here, sharing these musings in the still of the night, I can’t help but wonder: What’s the fate of serendipity in the age of AI?

If you enjoyed this dive into serendipity and scientific discovery, I’d love to hear your thoughts. Agree, disagree, or have a totally wild theory of your own? Let’s connect! Subscribe to my LinkedIn newsletter and let’s keep the conversation rolling.

READ | THINK | SHARE

From Atoms To Words

Today’s Story Filed Under: