AI Text Is Everywhere — Can Readers Still Tell It Apart?

AI now helps produce LinkedIn posts, e-books and roughly a third of web content. As machine writing gets harder to distinguish, detection is becoming a matter of verification, not instinct.

Why AI Text Is Harder to Spot Than Ever

Artificial intelligence has moved well beyond drafting emails. According to Forbes CZ's morning briefing, AI tools now routinely write LinkedIn posts, produce self-published e-books and help students and researchers with essays and papers. The digest cites an estimate that AI contributes to about a third of all web content, and notes that machine-written work has even won prizes in short-story categories.

The question the briefing raises is whether ordinary readers can still tell synthetic text apart from human writing. Its short answer is “yes, but”. The qualification matters: AI models have grown better at mimicking tone, structure and nuance, so the obvious cues that once gave them away are fading.

The same digest carries unrelated items, including Elon Musk's claim in an interview with The Economist that money will lose its meaning by 2036, and the record €30m transfer of Czech goalkeeper Lukáš Horníček to Newcastle United. The AI item is the one with the widest reach, because it affects anyone who reads, writes or publishes online.

The Detection Gap in AI-Generated Content

Why AI text is now hard to identify

The shift described in the briefing is not just about volume. As AI moves from obvious template text to polished, context-aware writing, detection becomes a probabilistic judgment rather than a simple scan. The source itself acknowledges this by answering “yes, but”: identification is still possible, but with significant caveats.

What the “third of web content” claim implies

The figure that AI contributes to about a third of all web content is striking, although the original briefing offers no methodology, so it should be treated as an estimate rather than a hard statistic. If accurate, it means much of what people read is generated or heavily assisted by machines. That shifts responsibility toward publishers to label automated material and verify what goes out under their names.

Where that leaves readers

The practical consequence is that surface style is no longer a reliable test. Readers and editors are left with slower checks: tracing claims to sources, comparing details across outlets and watching for text that has no independent reporting behind it. The briefing's “yes, but” is effectively an admission that spotting AI text now takes more work than it used to.

Checks for Readers and Publishers Facing AI Text

When a piece of writing cites no verifiable sources, treat its claims as unverified — AI already contributes to a large share of web content, so polished prose alone is no signal of reliability.
For editors and businesses, disclose where AI assisted in published material and keep a named human responsible for facts; the briefing shows AI is already mainstream in essays, social posts and e-books.
Readers should cross-check named facts, numbers and quotes against the original source, as with Musk's interview in The Economist, before drawing conclusions.