简体中文
Google didn't die. Chegg and Stack Overflow did — and nobody had named them.文章 · standmeet2026.10.06 · 文章
2026.10.06·2 分钟阅读#ai-predictions

The Ones That Died Weren't in the Predictions

Google didn't die. Chegg and Stack Overflow did — and nobody had named them.

What was said

From late 2022 through 2023, the hottest prophecy was "ChatGPT will kill Google": search would be replaced by answer engines, and a decade of dominance would fall. The Perplexities were written up as gravediggers. Meanwhile, almost no mainstream voice said the first to die would be a homework-answers website and a programmers' Q&A community — they weren't even granted the honour of being named.

What actually happened

Google didn't die. It folded AI summaries into its own results page, kept selling ads, and watched its stock hit new highs. The real collapses happened outside the spotlight, in two places. Chegg, a listed homework-answers subscription company, lost nearly half its stock value overnight when ChatGPT launched, then kept sliding — over ninety percent of its market value gone — its business model pulled up by the roots by a free chatbot. And Stack Overflow: traffic fell steeply and kept falling, as programmers stopped searching old threads and just asked the model; eighteen years of accumulated Q&A became, overnight, training data rather than a destination.

Why it didn't come true — and why nobody called it

Predictions love the biggest target, because "the giant falls" is the story. But disruption's real path starts from the undefended flank: homework answers and Q&A lookup are pure information-shuffling — no liability, no relationships, no ecosystem lock-in, zero switching cost — exactly what should have died first. Google has distribution, an ads system and the default entry point; for Google, AI is a feature upgrade, not a positional replacement. Who gets replaced depends not on how strong AI is, but on whether a business's moat happens to be nothing but information itself.

The unexpected part

Stack Overflow's ending has a bitter twist: its corpus fed the models, the models drained its traffic, and in the end it licensed its data to companies including OpenAI — from destination to supplier, a road many content businesses are now queueing to walk. The people who once predicted "content is king" did not foresee the king ending up living by selling territory.

就这篇文章向 AI 提问·上下文:“the ones that died weren't in the predictions”
›