AI promises to open up knowledge to everyone, like the Enlightenment did — but does it also quietly undo the fact-checking that made that knowledge trustworthy?

This explores whether AI continues the Enlightenment's push to open knowledge to everyone, or undoes it. The question is ambiguous, so here 'this' is read as AI. The corpus's sharpest answer is that it does both at once. Several notes draw on Adorno and Horkheimer's old argument that a technology built to free people can, by succeeding, create new forms of unfreedom. They argue AI does this to thinking itself Does AI repeat the Enlightenment's reversal into its opposite?. AI opens up access to answers on a scale Diderot could not have imagined. But the answers arrive in a form the Enlightenment spent centuries trying to get rid of.

That form is hearsay. The Enlightenment did not just make more knowledge available. It built tools for checking it: citation, archives, peer review, and chains of evidence that let you trace a claim back to its source. One note argues that AI output has every feature of pre-Enlightenment hearsay. It is testimony passed along secondhand, changed in each retelling, impossible to trace to a source, and impossible to check against a fixed original. So those verification tools can't process it at all Does AI-generated knowledge have the same structure as hearsay?. A companion note adds two more pre-modern traits: an appeal to authority the system hasn't earned, and a quiet sidelining of the reader's own judgment Does instrumental AI reproduce pre-Enlightenment knowledge structures?. In this reading, access goes up while the ability to check what you've accessed goes down.

A second line of argument is about form rather than truth. Print turned knowledge into a fixed stock you could accumulate, cite, and come back to. AI turns it back into a flow that is generated fresh each time, a bit like oral cultures. The difference is that it has no speaker or teacher whose presence once anchored that flow Is AI returning knowledge to flow-based economies?. Related notes describe the same loss from two other angles. AI claims multiply outside the social conversations that normally decide what counts as knowledge How does AI writing escape the conversations that govern knowledge?. And AI separates the finished product of thinking from the reasoning that would normally produce it Does AI separate intellectual form from the thinking behind it?.

The corpus doesn't only see regression. Venkatesh Rao argues that what makes LLMs new is not how much they cover, since the Encyclopédie already aimed to cover everything. It's that you can approach them from any angle and still get a coherent answer, where an encyclopedia or Wikipedia gives you one fixed way in Do LLMs succeed by being comprehensively encyclopedic?. That is a real expansion of access, especially for people who don't know the right vocabulary to look things up. Another note frames AI as humanity's pooled knowledge in condensed form. On that view, restricting who can use it would bring back the kind of gatekeeping the Enlightenment fought against Should restricting AI access create new kinds of inequality?.

The surprising part is that experiments suggest easier access can mean shallower learning. In seven randomized studies with more than 10,000 participants, people who learned a topic through ChatGPT summaries reported learning less and felt less ownership of what they learned. The advice they then wrote was sparser than advice from people who used ordinary web search Does learning from AI summaries produce shallower knowledge than web search?. The Enlightenment assumed that wider access would make people better at judging for themselves. The corpus suggests AI weakens that link: getting the answer becomes easier, and understanding it can become harder.