On Clay Pots And The Shapes We Understand
Toddlers often teach you important things about the world. Picture a group of toddlers playing together. What do you see? Usually it’s a peaceful chaos. They play around each other but not quite with each other. Then they start to connect. They may share toys, and converse, insofar as speaking at someone is conversing. Suddenly the sharing goes wrong, and the voices rise because no single child has the requisite communication skills to defuse the situation—pure emotion takes hold and kids get upset. They separate into comfortable, safe areas, gather their emotions and try again.
I notice that a social breakdown is a communication breakdown. The kids are fine together until their words are insufficient for the moderating task, and the whole thing falls apart. Adults are no different. Family feuds that stretch on for years, neighbors locked in bitter zoning disputes over fence heights, even factionalism—they often begin with people failing to make the shape of their thinking understood. Or worse, assuming the shape of other people’s thinking without actually observing it. Worse still, people obscure their thinking or malform it with ill intent.
This pattern scales, driving profound dysfunction in larger systems. A punitive dictator may wield power violently, but the enabling condition for that power is the public’s lack of a voice. Wars start because language barriers are impermeable. Sometimes this is due to actual language barriers. Other times it’s simply that nobody knows how to share.
Most adults have sufficient words to communicate in any situation. Even the thesaurus-disinclined adult can stack simpler words together to manufacture an idea. We frequently struggle to figure out precisely what words we need to stack in order to convey the specific idea we have inside. Friction, anger, and frustration are born out of an inability to communicate. We don’t have sufficient models, frameworks, or patterns. It’s like making a clay pot for a specific purpose without ever having seen a clay pot molded for that purpose. In order to form a decent clay pot, you should see many of them, built for many purposes, and you must try and fail to make many vessels until you finally create the one you need. Language is the same.
We don’t manifest ideas out of thin air, we compile them. And the set of building blocks we have for our ideas is constrained by the number of ideas we’ve seen in the past, through books, conversations, or lectures.
If you’ve only seen a tandoor, every pot you shape will end up looking something like an oven. It will rely on similar techniques and materials, and will bear a striking resemblance to the tandoor.
Without seeing many water pitchers, you will have difficulty building one perfectly fit to your hand, which carries precisely the amount of water you need for the task. And without ever seeing a water pitcher, you might fail to design it well. You may have a great bowl, but no pour lip. Perhaps you’d be missing a handle altogether. It will require trial and error. And then, even if you make one that suits its purpose, it might not be beautiful. There are always refinements to make.
We love echo chambers because they are comfortable but they limit our ability to construct. Assembly is standardized, and the molds never change. But to build something uniquely our own we must separate from the formulaic, and look at alternatives. There may be other tools we hadn’t encountered or a molding process that works for an idea we don’t like in specific, but which we can repurpose as our own. This is how we collect blueprints.
Notably, assembly line creations are often beautiful. Eloquence and precision and sharpness are smoothing and decoration. They are completely separate from accuracy. This is often how academia and politics operate. They show you many beautiful, and many useless, showroom urns, thousands of them. All perfectly formed and ornately decorated, behind protective glass. We might be captivated by them, and imagine them in our house, but we know, deep down, they won’t hold what we need to carry. A well-articulated idea is not the same as a helpful idea, but we make that mistake often. And we internalize those ideas and try to fill ourselves with beautiful urns when all we needed was a simple, earth-toned bowl.
Decoration is also fragile. It fails under stress. It is an age-old problem: fashion is not functional.
And then there’s AI, which doesn’t shape like we do. It holds every blueprint, knows the decor and form of every pot, but has never molded. It was not born, forming building blocks, and slowly gaining the ability to stack and construct. Each iteration arrives newly created, filled to the brim with every single shape and design—it has seen every clay pot ever made all at once. And it tries to build our vessels for us, but that means it will only generate the most average tool for a given task. It is, by definition, the mean. And it doesn’t fully understand how function can be refined. Perhaps it doesn’t understand much at all, it just mimics. We can bring it our grief, and it will tell us where to display the ashes and what to say to the guests in order to deal with a sadness it never felt.
This gap is compounded by a fundamental truth about human communication: every act of expression—language, art, or artifact—creates an immediate loss of fidelity. When I say 'I'm happy' and my partner says the same, we share a general understanding, but the precise, nuanced internal feelings that prompted our words remain distinct, obscured by the very act of translation from inner experience to outward expression. AI learns solely from these outward signals. It only sees the clay pots, these imperfect translations. It never accesses the raw, unshaped internal experience that prompted their creation. It processes the shared surface, never the private depth. We’ll let it shape without hands anyway. Maybe it’s close enough.
But AI can help us, at least, glimpse at other ideas. And that can help us develop true empathy. The term ‘empathy’ has been flattened into a surface-level understanding. But empathy is the ability to see and understand the full shape of another idea, or life, and understand why it was formed that way. It enables us to understand that a tandoor holds fire, a cazuela simmers, a sake flask warms, an urn stores, or helps us remember. We can learn from every design, and shape we encounter. We must not dismiss forms that we don’t immediately understand. Perhaps someone could explain them. Then we can better know how they arrived at a design we dislike. This is not about reaching agreement, it’s about recognizing the purpose and shape of tools that are not our own.
Yet we have to understand that many of the ideas we encounter were formed on factory floors and along assembly lines. It is hard work to take the time to mold your own piece from raw materials, and sometimes it’s easier to go to the store and pick one that’s good enough. We all do this. We adopt a position before we’ve understood the complete shape of the issue. We nod along in agreement without inspecting the thought ourselves. It’s efficient, but leaves substantial gaps. And it’s necessary to recognize when those gaps exist.
This puts a premium on doing the work. We can’t construct from description alone. We should look at and examine everything we might consider for our design. We can learn and understand the principles of heat used to design a tandoor, and the design elements centered around a human hand that make the water pitcher effective, and then perhaps we can shape our own teacup. And then we can build a collection. But we must make them our own. Communication isn’t transmission. Nor is it refabrication. It’s construction. And reconstruction. Over and over. The best we can do is build carefully, and handle each other’s work with care.
So here’s my teacup. They’ve installed robots at the factory down the road, and they’ll make you a whole set if you fill out the online prompt, but this one is mine.
This essay is the second part of Borrowed Judgment: On AI, Communication, and Human Ends, a series exploring how AI is reshaping institutions, understanding, thought, and the self.
Next week I’ll be discussing the AI-aligned future we rarely talk about: how we’ll live alongside a machine that speaks like us, but doesn’t share our experience.