A reply to Matt Johnson’s “Reading Orwell in the Age of AI.”

The real risk here isn’t just that AI will produce inauthentic writing. It’s that it may detach writing from the experience of judgment under constraint. Orwell didn’t just write well—he learned to think by confronting situations where something had to be decided and someone had to bear the cost. That’s what gives his prose its authority.

AI can replicate language, structure, even moral tone. But it cannot participate in the process by which a writer comes to accept responsibility for what is being said—who it affects, what it implies, what it costs. If that link breaks, we don’t just lose individuality in writing. We risk producing a culture fluent in moral language but increasingly unable to sustain the burden of judgment that gives that language meaning.

And there are already signs of this shift. In universities, media, and professional discourse, we often see high-confidence moral claims paired with a striking absence of engagement with trade-offs, consequences, or cost-bearing. That may not be caused by AI—but a world in which writing is increasingly detached from lived experience and responsibility will almost certainly amplify it.

—Allen Zeesman