In East Germany, correspondence chess was hugely popular. The internet was decades away, and a private telephone line was a rarity. Anyone who wanted to test their skill wrote their move—say, 1. e2–e4—on a postcard, dropped it in the mailbox, and waited days for a reply. At home stood a wooden board where you patiently recreated your distant opponent's moves and watched the game unfold over weeks and months.
My father once told me about a trick as simple as it was brilliant. Someone started two correspondence games at once against two strong players—playing Black in one and White in the other. When the first player's opening move arrived, the person simply copied it onto a new postcard and sent it to the second player as their own. The reply went straight back to the first. Without the slightest idea, the two players spent months playing against each other, while the person in the middle paid the postage and watched two distant minds compete on their board.
Today I use this principle almost daily—with one decisive twist. On my screen run two of the most capable language models of our time: Claude and Gemini. When I'm refining a delicate letter to a public authority or rebuilding the architecture of this website, I pass one AI's answer straight to the other.
I ask Claude for a draft and show it to Gemini. Gemini praises the structure, spots a pitfall, and rewrites it. I hand that version back to Claude—and Claude dryly points out where Gemini overreached, missed details, or struck an overly dramatic tone. The next move is the fascinating one: back with Gemini, the model doesn't defend its turf but replies, in essence: “That critique hits the mark—I overdid it in two places.”
The old chess trick has become something new. The two systems aren't playing blindly against each other. They see each other's moves openly on the table and pass the ball back and forth. Technically, this dialogue could easily be automated. Yet in my role as the human courier between two different machine logics, a productive friction emerges that yields wiser results than either system alone.
Watching this ping-pong raises a question of justice: where does this seemingly superhuman intelligence come from?
An artificial intelligence is as smart as the data it was trained on. And that data is text. Billions of sentences carry the ideas, errors, poems, scientific breakthroughs, suffering, and hope of thousands of years of human history. None of that knowledge was created by Elon Musk, Larry Page, Sergey Brin, Jeff Bezos, or Peter Thiel. And yet its yield accumulates in a handful of corporations—and with men whose personal fortunes have grown to hundreds of billions of dollars, sums beyond anything a single life could meaningfully use.
The irony isn't lost on me: Gemini belongs to Google, and Claude is backed by multibillion-dollar investments from Amazon and Google. My cooperative chess is played on boards owned by the very companies I'm describing. That makes the question of who owns the game all the more urgent.
If the raw material of artificial intelligence is the shared cultural heritage of our species, its harvest cannot remain the private monopoly of a few companies. We need a global framework for an AI Commons: fair access for all of humanity to the tools distilled from our collective memory.
Right now, the digital economy is moving in the opposite direction. Instead of an open commons, we are seeing the privatization of human conversation.
There are early attempts to rebalance things. Cloudflare, for instance, is testing pay-per-crawl, a model in which AI crawlers pay for access to websites. At best, that is a transition. Meanwhile, platforms are selling their users' conversations: Reddit has licensed its data to Google for AI training, reportedly for around $60 million a year, and has a similar deal with OpenAI. And from mid-December 2026, Discord is making messages from more than 63,000 public servers findable by search engines—conversations that until now largely stayed inside the platform.
What gets discussed on Reddit and Discord is the raw pulse of our present: unfinished thoughts, doubts, subcultures, collective searching. Understanding where we are heading as a society should be an open good for everyone—not exclusive raw material for the training pipelines of a few trillion-dollar corporations.
What happens to us when we realize that every word we type in public is vacuumed up, monetized, and locked inside closed models? We withdraw. Few of us will keep entrusting our most vulnerable ideas to a faceless crawler.
This is where the global question of the AI Commons meets the street outside our door—and the idea of the neighborhood reporter (Kiez-Reporterin). That is exactly why we at UNO.TEAM send out a real neighborhood reporter instead of an algorithm: a person who meets people in our Leipzig neighborhood face to face, right in the café or the shop, and listens to them. Nobody opens their heart to a crawler. But to a person sitting in the café on the corner, someone who listens and with whom trust grows over weeks and months—to them, we open up. We tell them what really moves us: the story behind our shop, our escape, or our idea.
Perhaps that is the dual strategy for the years ahead. On one side, we let the great AI models work together like correspondence chess partners and reclaim them as humanity's shared inheritance. On the other, we answer the crawlers with something no data center in the world can copy: a conversation between people at a table in our own neighborhood.
Three questions to keep thinking. Choose an answer – you will then get an explanation and a prompt for self-reflection.
Who would you like to talk to about this thought?
Thank you. Every answer fits – take it with you into your day. Would you like to share something yourself?
