What if Luther hadn't had the printing press?
Ideas don't always change the world. Often it's the paths they travel.
Imagine Martin Luther completing his Bible translation in a world without a printing press. The texts would have remained handwritten, slowly multiplied, carried from town to town, and read by only a few. There would have been unrest and dispute even in this case, but very likely only in individual communities. It would hardly have grown beyond a rebellious theological event. That fire, which would have unhinged an era, would have fizzled out at the spark.
Now the inverse picture: The press is ready, the type is cast, but no one dares to question the prevailing order of knowledge. There would be machines, but no sentence that burns; an infrastructure, but no translation that wrests the monopoly of interpretation from the authorities and calls out to the reader that they no longer have to wait for someone to explain what is true. We would have the technology. But no liberation.
The historical point lies precisely in this intermediate space. Neither Luther nor the printing press changed the world alone; it was their confluence. A new interpretation that met a new infrastructure. One posed the question differently, the other ensured that it could never again be contained, and from both emerged a third.
Man could relate himself to truth.
But this did not happen smoothly, and certainly not overnight. It cost violence, error, power struggles, and in the end, new dogmas stood where old ones had fallen. History rarely proceeds neatly. The door that now stood open empowered those who mastered reading. Thus, the interpretation of the word no longer rested solely with the few.
Today, we stand at a similar threshold, but under inverse signs. The question is no longer whether texts can be duplicated. It has shifted. Who is allowed to produce knowledge at all? Large language models don't just output texts. They organize arguments, formulate hypotheses, dissect data, design strategies, write programs, compare studies, simulate decisions.
They are not libraries that preserve existing things, nor are they printing presses in the old sense that merely multiply existing things; they are involved in the creation of knowledge itself. And this is where it gets uncomfortable. Because as long as knowledge was created in universities, editorial offices, laboratories, and public institutions, it was at least broadly visible where it came from. If it now shifts into closed model systems, the old power question returns in new guise. It once asked: Who owns the books? Then: Who controls the media? Today, increasingly: Who owns the systems that produce new answers? Perhaps therein lies the true parallel to the Reformation. Not in religion, church, in Luther as a hero or heretic, but in the question of what happens to a society that relinquishes its tools of knowledge.
So let's think the scenario through. There are powerful AI systems, but only a few understand how they work. That would be the printing press without the art of reading. The machine runs, the pages fly, the world shifts, and most stand by, seeing only what comes out in the end. They can admire the sight. They cannot devour, fear, criticize, or reject the content.
The picture is sharpened as soon as these systems no longer merely reproduce existing things, but produce new ones. Despite the many critics who warn that today's AI is only a very successful sleight of hand of mathematics, it must still be admitted that even today's systems, in the hands of capable people, are significantly involved in the creation of new knowledge. Undiscovered patterns, unseen solutions, fresh scientific conjectures, market strategies, political narratives can also emerge through skillful questioning and feeding of the systems. Anyone who sees only mere access as the solution here, I must disappoint, because this is no longer sufficient. What becomes crucial is not whether one receives an answer, but whether one understands how it came about. Whether what is read is probably true or merely sounds convincing stands as the basis of the new fragmentation. Its first layer is openly apparent. Is it knowledge that stands before one, or merely polished language? This is more dangerous than mere inequality – it is a new form of epistemic dependence. If a few actors determine which models are built, which data are counted, which answers seem plausible, and which questions are not even asked, a classic knowledge monopoly with visible walls does not arise, but something much finer, softer, harder to grasp.
Today, we stand at a similar threshold, but under inverse signs. The question is no longer whether texts can be duplicated. It has shifted. Who is allowed to produce knowledge at all? Large language models don't just output texts. They organize arguments, formulate hypotheses, dissect data, design strategies, write programs, compare studies, simulate decisions.
They are not libraries that preserve existing things, nor are they printing presses in the old sense that merely multiply existing things; they are involved in the creation of knowledge itself. And this is where it gets uncomfortable. Because as long as knowledge was created in universities, editorial offices, laboratories, and public institutions, it was at least broadly visible where it came from. If it now shifts into closed model systems, the old power question returns in new guise. It once asked: Who owns the books? Then: Who controls the media? Today, increasingly: Who owns the systems that produce new answers? Perhaps therein lies the true parallel to the Reformation. Not in religion, church, in Luther as a hero or heretic, but in the question of what happens to a society that relinquishes its tools of knowledge.
So let's think the scenario through. There are powerful AI systems, but only a few understand how they work. That would be the printing press without the art of reading. The machine runs, the pages fly, the world shifts, and most stand by, seeing only what comes out in the end. They can admire the sight. They cannot devour, fear, criticize, or reject the content.
The picture is sharpened as soon as these systems no longer merely reproduce existing things, but produce new ones. Despite the many critics who warn that today's AI is only a very successful sleight of hand of mathematics, it must still be admitted that even today's systems, in the hands of capable people, are significantly involved in the creation of new knowledge. Undiscovered patterns, unseen solutions, fresh scientific conjectures, market strategies, political narratives can also emerge through skillful questioning and feeding of the systems. Anyone who sees only mere access as the solution here, I must disappoint, because this is no longer sufficient. What becomes crucial is not whether one receives an answer, but whether one understands how it came about. Whether what is read is probably true or merely sounds convincing stands as the basis of the new fragmentation. Its first layer is openly apparent. Is it knowledge that stands before one, or merely polished language? This is more dangerous than mere inequality – it is a new form of epistemic dependence. If a few actors determine which models are built, which data are counted, which answers seem plausible, and which questions are not even asked, a classic knowledge monopoly with visible walls does not arise, but something much finer, softer, harder to grasp.
A monopoly on probabilities, on interpretation, on what sounds reasonable.
And one might not even notice it. Precisely because the answers are so friendly, so fluid, helpful, and self-assured, that no feeling of exclusion arises. One does get an answer. But an answer is not yet participation. Reading was once more than a craft; it was a shift in power, for whoever could read no longer had to believe what was read to them.
The corresponding art of our time is not exhausted in prompting, and certainly not in a few tricks for dealing with chatbots. It consists of generating, testing, and contradicting knowledge with AI. This requires an understanding of how such systems work. It requires a creative streak to understand their actual utility. It requires everyone to decide whether they consume the product, whether the product consumes them, or whether they participate in the product as producers. Then AI will not become the new pulpit from which pronouncements are made, but rather the workshop in which work is done.
Perhaps that is the real lesson of the old thought experiment: A society does not become free because a technology exists, nor because individual people have great ideas. Freedom only arises where idea, tool, and ability come together. Luther without the printing press would probably have faded away. The printing press without readers would have liberated little. A powerful AI without broad AI literacy may become many things – it will not become democratic on its own
The corresponding art of our time is not exhausted in prompting, and certainly not in a few tricks for dealing with chatbots. It consists of generating, testing, and contradicting knowledge with AI. This requires an understanding of how such systems work. It requires a creative streak to understand their actual utility. It requires everyone to decide whether they consume the product, whether the product consumes them, or whether they participate in the product as producers. Then AI will not become the new pulpit from which pronouncements are made, but rather the workshop in which work is done.
Perhaps that is the real lesson of the old thought experiment: A society does not become free because a technology exists, nor because individual people have great ideas. Freedom only arises where idea, tool, and ability come together. Luther without the printing press would probably have faded away. The printing press without readers would have liberated little. A powerful AI without broad AI literacy may become many things – it will not become democratic on its own
For whoever only reads what others produce, remains a spectator.
But whoever understands how knowledge is created, can object.
And sometimes, freedom begins precisely there.
But whoever understands how knowledge is created, can object.
And sometimes, freedom begins precisely there.