Curiosity and Serendipity: Our Edge in the Age of AI
An Accidental Discovery and a Thought Experiment
In 1928, a Scottish scientist named Alexander Fleming made a serendipitous discovery that would change the course of medicine. Upon returning from a summer holiday, Fleming noticed that one of his petri dishes containing Staphylococcus bacteria had been contaminated with a fungus. Curiously, the area around the fungus was clear of bacteria, suggesting that the mold had secreted something that inhibited bacterial growth.
Intrigued by this unexpected observation, Fleming’s curiosity compelled him to investigate further. He identified the mold as a rare strain of Penicillium notatum and speculated that it produced a bacteria-killing substance. Fleming named this substance “penicillin” and published his findings in 1929.
However, it would take over a decade and a team of Oxford scientists to purify penicillin and demonstrate its incredible potential as an antibiotic. In 1945, Fleming, along with Howard Florey and Ernst Chain, was awarded the Nobel Prize in Medicine for this groundbreaking discovery that has since saved countless lives.
The story of penicillin’s discovery is a testament to the power of human curiosity and serendipity in driving innovation. Fleming’s chance observation and his relentless pursuit to understand it led to one of the most significant medical breakthroughs of the 20th century.
But could an artificial intelligence like ChatGPT have made this discovery?
First, a Fundamental Question — Is AI creative?
Is AI creative? Can it innovate? That’s the topic of a raging debate happening across academic and business circles right now. The answer has tremendous implications for the Future of Business, the Future of Work and the very foundations of what it means to be human.
The answer is both “Yes” and “No”. LLMs, the vanguard of the latest revolution in AI, have tremendous ability to identify hidden patterns among human knowledge. Is this creativity? The Godfathers of AI Ray Kurzweil and Geoffrey Hinton certainly think so.
Yet creativity is a broad-term with many buckets.
Going back to my thought experiment. I do not think that AI could have replicated the discovery of Penicillin, simply because it lacks the uniquely human qualities of open-ended curiosity and serendipity that were so crucial to Fleming’s success.
#1 AI lacks true curiosity
AI cannot replicate true curiosity. In (a dated but relevant) article from the Harvard Business Review, the authors argued that while AI may exhibit “task-specific curiosity” within set boundaries, the uniquely human trait of open-ended and spontaneous curiosity may be something that AI cannot fully replicate. This is because current AI is goal-oriented and operates within boundaries set by its training data.
While AI may develop limited forms of “artificial curiosity”, most experts remain skeptical that this can reach the depth and spontaneity of human curiosity in the foreseeable future. First, human curiosity emerges from innate qualities like self-awareness, emotional depth, and the ability to find deeper meaning. In contrast, today’s AI systems lack these core traits that shape human inquiry. LLMs like GPT also simply mimic human-like outputs based on statistical patterns without deeper contextual understanding. Crucially, these systems cannot yet replicate the key drivers of human curiosity like noticing and investigating unexpected observations, imaginatively linking disparate ideas, and directing their own learning through an innate sense of curiosity.
#2 AI cannot replicate human serendipity
The rich diversity of over 8 billion human lived experiences across cultures represents a vast well of serendipitous insights that are inaccessible to AI systems. Our unique personal contexts, from the sights and smells we encounter to the formative events that shape our perspectives, provide the raw material from which humans can derive creative ideas and make unexpected conceptual connections. AI, in contrast, is limited to the data it is exposed to during its training.
Yes, AI may develop increasingly sophisticated ways to perceive and interact with the world through sensors and embodied systems, but the prevailing view is that replicating the depth and richness of human lived experience remains a profound challenge due to fundamental differences between human and machine cognition. Crucially, human experience is deeply embodied — our minds are inextricably intertwined with the sensory modalities and physicality of our bodies as we interact with our environment. This embodied existence is central to developing the emotions, personal contexts, and formative experiences that fuel our curiosity and serendipity.
With AI, the boundary for value creation shifts from problem solutioning to problem discovery / definition
AI is already transforming innovation by enabling new discoveries and shortening innovation cycles. As these models become more powerful, they will continue to turbo-charge problem solving in unprecedented ways. Yet, their goal-oriented nature and necessarily incomplete training data limit their innovative capabilities.
As AI gets deployed at scale, the boundary of value creation will shift from just problem solving to problem definition and discovery. Asking the right questions, noticing the unexpected, providing essential context and setting the right innovation agenda for AI to explore will be increasingly critical.
Engineering Curiosity and Serendipity: a New Organizational Imperative
What then can companies do to stay competitive amid the tumult and promise of AI-driven innovation? Here’re some suggestions
- Hire for curiosity. In a world where AI can quickly acquire knowledge, raw intelligence matters less than a person’s ability to learn. Hire for learning ability rather than experience. Seek out candidates who ask insightful questions, have wide-ranging interests, and demonstrate a growth mindset. Look for those who have taught themselves new skills.
- Build a learning organization. Create an environment that encourages continuous learning, experimentation, and knowledge sharing. Provide opportunities for training and development. Emphasize learning goals over performance goals. Make it safe to take risks and learn from failures.
- Engineer serendipity. Design workspaces to facilitate chance encounters and casual conversations across teams, like Pixar’s central atrium. Create coworking spaces and communal areas where employees from different departments can interact. Hold cross-functional events and encourage people to build relationships and share ideas outside their immediate teams.
- Carve out space for exploration. Give employees dedicated time, like Google’s famous “20% time”, to pursue self-directed, open-ended projects. Allow people to follow their curiosity down unexpected paths without worrying about failure.
- Design bionic, human-AI processes. Rather than having AI take over entire workflows, thoughtfully design processes to include opportunities for human intuition, judgment, and exploration at each stage. Have people frame the problems for AI to work on. Encourage employees to approach AI-generated insights and solutions with curiosity — to ask questions, spot potential issues or novel opportunities, and redirect AI toward promising new paths it might have missed.
Ultimately, while AI will continue to push the boundaries of what’s possible, the human capacity for open-ended curiosity and serendipity will continue to make human insight and ingenuity valuable. Our ability to notice the unexpected, draw meaning from experience, and imaginatively connect ideas in original ways has driven transformative breakthroughs throughout history — think microwave, X-rays, Velcro, Vulcanized rubber — and will continue to do so in future.
That is not to say that AI is inferior to human innovation or vice versa. Rather, the most powerful innovations of the future will likely emerge from a collaboration between human and machine intelligence, not from AI alone.
Stay Sharp and Snappy,
The Snappy Strategist
