🚨 THE AI “TAKEOVER” GRAPH IS MISLEADING

Yes, LLMs are generating massive token volume. But tokens are not the same as meaningful written output, and treating them that way distorts the entire picture.
Most AI text is functional, not enduring. Code completions, chatbot replies, draft iterations, support scripts. High volume, low permanence, often never read twice.
Meanwhile, human output is massively undercounted. The “200 trillion words” ignores messages, emails, internal docs, and the constant stream of informal writing that actually drives communication.
So the comparison breaks. We are measuring curated human publishing against raw machine exhaust, which makes the curve look far more dramatic than reality.
THIS IS THE REAL STORY
AI is scaling production, not replacing judgment. It expands drafts, accelerates workflows, and increases surface area, but meaning, taste, and impact still come from humans.
The shift isn’t AI surpassing us. It is humans writing with leverage, and markets misunderstanding what is actually being measured.