Simon Says…Entropic Knowledge
Over the past few years, my colleagues and I have been deeply immersed in projects applying AI to seemingly intractable problems. We've faced countless challenges—some quite exasperating—but the power of the tools we've built continues to astonish us. Every few weeks, I find myself grappling with an existential question: "What will be left for humans to do?"
The Limitations of Omniscient AI
Recently, I pondered this question in a more tangible form: "Would I call Simon if I had an all-knowing AI assistant?" By "omniscient," I mean an AI possessing all available information, excluding the ability to predict the future or read minds. Simon is Simon Fowler, my brilliant, insightful, and compassionate friend whom I call whenever I stumble upon something intriguing to me—usually with complete disregard for his own interest in the topic.
My immediate answer was, "Of course, I'd still call Simon!" But then I asked myself, "Why would I?" If I have access to an AI that knows everything, what value does Simon add? (Simon asked me the same thing when I called him.)
The Uniqueness of Human Knowledge
The personal connection is obvious. I enjoy hearing about Simon's family, his latest adventures, and his charitable work. But what about the brainstorming sessions—the mental laboratory that our conversations create? Why would I need that if I had a 24/7 omniscient AI?
As I reflected on the countless conversations I've had with friends and colleagues, I realized that while their breadth of knowledge is valuable, it's actually the limits of their knowledge and how they've acquired it that offer the greatest value.
Each of us collects an immense amount of knowledge across our lives. Much of that knowledge we acquire intentionally whether out of curiosity, passion, or a desire to improve our capabilities. We only retain a small portion of the knowledge we encounter (which is a tiny fraction of what is knowable) and what we retain is an imperfect reflection of the objective truth, shaped by our previously encoded knowledge, what our body is sensing at that moment, and our emotions. This rich form of information we retain is sometimes called sensor-affective knowledge. Consciously and unconsciously, Simon curates his sensory-affective knowledge library, connecting it in novel ways to the other entries grown into his brain. Simon’s perspective is shaped by his unique journey, as much by what he chose not to retain as what he did, and the texture, temperature, tenure, tone, and tint of that knowledge. (Sensory-affective knowledge makes me think of the “Your move, chief” scene in Good Will Hunting.)
Defining Entropic Knowledge
In Information Theory, entropy is a measure of uncertainty or unpredictability in information content. For Wordle fans, entropy helps determine the optimal next guess—the word that will provide the most information about the hidden word and minimize the expected number of guesses.
This got me thinking about the tacit knowledge people possess, which I categorize into three buckets:
领英推荐
I call the valuable part of this third category "Entropic Knowledge.” (Not all of it is valuable, and none of it is valuable in all situations.) Entropic Knowledge represents the unique perspectives and insights individuals bring to discussions, brainstorming sessions, or problem-solving processes. These insights stem from personal experiences, intuitions, and internalized understandings that are difficult to articulate or replicate. By influencing thought patterns and perspectives, this form of knowledge increases the likelihood of finding innovative solutions or novel approaches. Originating from personal experiences, intuitions, and internalized understandings that cannot be easily codified or simulated, entropic knowledge introduces a high degree of creative potential into collaborative problem-solving efforts, fundamentally altering cognitive interactions between two or more people.
The Irreplaceable Value of Human Collaboration
What Simon focuses on when I speak, the questions he asks, the analogies he offers, and the stories he shares are unique to him. His perspective is shaped by his life's journey—by what he's chosen to retain and how he interprets that knowledge. This uniqueness influences how I think about problems or opportunities, introducing a high degree of variability and creative potential into our collaborative efforts. Simply put, I think better because Simon brings a lot of Entropic Knowledge to our conversations.
This isn't exclusive to Simon. Other friends and colleagues provide similar value, each bringing their unique Entropic Knowledge to the table. Their contributions are context-specific, hinging on the nature of our relationship and the issue at hand. The familiar ways our conversations unfold allow the potential of their Entropic Knowledge to be fully realized for me.?
Why Humans Are Irreplaceable
An omniscient AI would undoubtedly be incredibly valuable and capable of things humans can't do. I would certainly make extensive use of such a tool. However, it would complement rather than substitute the unique insights my friends provide. The AI's contributions would reflect its own way of acquiring and encoding knowledge. Even if it could replicate Simon's neural pathways at a given moment, it would diverge from the real Simon from that point forward.
The AI would bring its own Entropic Knowledge, but it wouldn't replace the unique contributions of individuals like Simon. Their irreplaceable value lies in the unpredictable, deeply personal insights that only come from their life journeys and the sensory-affective knowledge they have curated.
Conclusion
So, would I still call Simon even if I had access to an omniscient AI? Absolutely. Because the richness of human connection and the unique, unclonable insights that come from our shared experiences are things that no AI can replicate. In an increasingly AI-driven world, it's these human elements that will continue to inspire innovation and deepen our understanding.
Codifiable knowledge has been codified for years and recent technological developments are accelerating the process. Simulatable knowledge will be commoditized too through Advanced Contextualization Engines . Entropic Knowledge, however, should remain uncommoditizable even if they find a way to upload our minds and souls into machines.
A Note to Simon
Rest easy, Simon (and Chris, Dave, Ellen, John, Kata, Marcia, Mark, and Robert)—I will continue to interrupt your lives with my calls! After all, some things are simply irreplaceable.