I Understand the Aversion. I Just Don’t Think It Leads Anywhere Good
On cave paintings, echo chambers, and who really benefits from the pro-contra war over AI.
🔵 AGI Transformation / Agentic AI · 🔴 Existential Risk · 🟠 Present Harms · 🟢 Governance · 🟣 Sovereign AI · 🩵 AI for Science
© 2026 Björn Burkhard · CC BY 4.0 · bjornburkhard.com/ai-discourse-landscape-2026
I understand why people recoil when they hear the words “generative AI.” I understand the feeling that something has been taken — that the texture of creative work, of language, of image-making, of the space between human intention and human output has been smoothed over by a machine that didn’t earn it. I understand the sense that something fake has crept in, dressed in the clothes of something real.
That feeling is not nothing. It deserves to be taken seriously.
What I’m less sure about is where it leads.
The authentic and the “authentic”
There’s a version of the aversion that has a coherent logic. It goes something like this: generative AI produces outputs that mimic the surface of human expression without the experience that generates it. A poem written by a model carries none of the weight of the life that — in a human writer — would have made the poem necessary. The result looks like the thing, but the process that gave rise to it is fundamentally different.
That’s a real distinction. Worth making. Worth defending.
But the argument has a tendency to slide somewhere else. It slides toward the claim that authentic expression has always required a particular form — a particular material, a particular struggle, a particular kind of slowness. And once it slides there, it starts to look less like a defense of something real and more like a defense of a specific historical moment in the technology of expression.
The wax tablet was the generative AI of its time. So was the printing press. So was the typewriter, the synthesizer, the digital camera. Each time a new tool arrived that lowered the barrier to a form of expression, someone argued that the lowering itself was the corruption. That what made the thing meaningful was precisely the difficulty of producing it.
Sometimes they were right. More often, they were mourning a craft practice while the culture moved through the door anyway, taking with it everything that mattered about the original impulse.
The cave painting is not more real because it was made with pigment and a hand on cold stone. It is real because a human being made it, for reasons, in a world they were trying to understand. The medium carried the intention. The intention was not the medium.
The bubble that feels like a position
Here is what concerns me more than any argument about AI’s aesthetic legitimacy.
We are living through a period in which human discourse — actual exchange between people with different experiences, different assumptions, different conclusions — is becoming structurally harder. Not because people are stupider or less curious. Because the infrastructure of how we encounter ideas has been tuned to return us to ourselves. Recommendation algorithms, social platforms, search engines optimized for engagement: each of these, independently and in combination, has the effect of narrowing the range of things that feel plausible to us.
The result is not a culture of debate. It is a culture of camps. And the camps are not arguing. They are performing their position to an audience that already agrees with them, at increasing volume, with decreasing actual contact with the other position’s best version of itself.
The “pro AI versus contra AI” split is one of the places this is most visible.
On one side: an enthusiasm that can tip into evangelism, where every technology is a solution, every hesitation is fear, every critique is a failure of imagination. On the other: a resistance that can tip into refusal, where the appropriate response to a genuinely complex tool is a clean rejection that requires no further engagement.
Both positions have one thing in common: they are very easy to hold. They require no new thinking. They supply a ready-made community of people who will reinforce the position rather than test it. And they guarantee that whatever is actually happening — the real shifts in how people work, create, communicate, understand — stays at a safe remove.
Get Björn Burkhard’s stories in your inbox
Join Medium for free to get updates from this writer.
The cave painting argument is not a conversation. It’s a retreat.
© 2026 Björn Burkhard · CC BY 4.0 · bjornburkhard.com/ai-discourse-landscape-2026
Who benefits from the war
It’s worth asking a simple question: if we stay locked in a permanent pro-contra war about generative AI, who benefits?
Not the people who are genuinely trying to work out what this technology means for their practice, their work, their creative life. Those people need nuance, examples, friction, genuine disagreement with someone who’s thought carefully about the other side. The binary doesn’t give them that.
Not the culture. A culture capable of absorbing a genuinely transformative technology and making something of it on its own terms needs the kind of discourse that can hold complexity — that can say “this is useful” and “this is dangerous” and “this raises a question I don’t know how to answer” in the same sentence.
The binary benefits, above all, the entities that want to extract value from human attention and human reaction. The ones who need engagement, clicks, outrage, tribal affiliation. The ones who profit from the fact that you never actually have to encounter the other side’s strongest argument, because the algorithm has already sorted you into a bucket and the content will keep you there.
The human spirit — the actual capacity for encounter, for exchange, for being changed by contact with something outside yourself — is the casualty of that arrangement. And it is worth noting: that casualty is not AI’s fault. It is the fault of a discourse infrastructure that preceded AI and will outlast it if we don’t pay attention.
What AI can actually do to a bubble
Here is something I’ve noticed in practice.
A well-used AI tool — one you actually push back against, one you ask hard questions of, one you deliberately instruct to argue the opposite of what you believe — can function as a pressure test for your own position. Not because the model is right, or because it has a view. But because it can articulate, on demand, the best version of an argument you’ve spent no time inside. It can give you the friction that your feed has been carefully removing.
This is not a utopian claim. It can also reinforce what you already think, if you use it that way. Every tool follows the intention behind it.
But the capacity is there. And it points at something the aversion misses: the question is never only what the tool is, but what we make it do.
A cave painting is not a position statement. It’s a record of someone paying attention to the world and trying to make that attention visible. Generative AI, at its best, can be an instrument for that same kind of attention — directed outward, toward others, toward ideas that make you uncomfortable, toward the parts of the world your bubble has been editing out.
What we owe each other
The thing the pro-contra war costs us most is not an argument. It’s a practice.
The practice of actually engaging with someone who has thought carefully about a position you find wrong. The practice of staying in the conversation past the point where it gets uncomfortable. The practice of updating — not capitulating, updating — when something lands that you didn’t expect.
That practice is not a byproduct of having the right tools. It requires a decision. A decision to treat the person across from you — or the idea across from you — as something worth the difficulty of genuine contact.
I understand the aversion to generative AI. In many forms it reflects something real: a sense that the cost of things matters, that the difficulty of making is not separate from the meaning of what’s made, that speed and abundance are not the only values worth optimizing for.
I hold some version of that myself.
But aversion hardened into a position, insulated from friction, circulated in a community that never tests it — that’s not a defense of anything. That’s just another bubble. And bubbles, whatever their content, benefit the same people.
The human spirit doesn’t need protection from tools. It needs protection from arrangements that keep it from ever actually encountering anything.
That’s the conversation worth having.
