The 25-year internet bet and lessons for the AI debate today
A $1,000 bet in 1995
In March 1995, Kevin Kelly — executive editor of Wired magazine and a believer in the transformative power of technology — visited Kirkpatrick Sale's apartment in Greenwich Village, New York. Sale was a historian who had just published Rebels Against the Future: The Luddites and Their War on the Industrial Revolution, in which he called the internet and computers a deadly danger to the planet.
The interview quickly turned into a shouting match. Sale argued that digital technology would destroy jobs, push millions out of the economy, widen inequality to the point of a rich-poor civil war, and that the environment would collapse. Kelly held the opposite — the internet would create new professions we hadn't yet imagined, expand markets, and create wealth for more people than ever before.
At the end of the interview, Kelly pulled out his checkbook, wrote $1,000, and said: "Then let's bet. By 2020, if the world is as you describe — global currency collapse, a rich-poor war, environmental catastrophe — I lose. If not, you lose."
Sale signed. The bet became one of the most famous wagers in the history of technology.
The result: 25 years later
In 2021, the intermediary editor who held the check — Bill Patrick — announced the result. Kelly won. The internet did not destroy society. On the contrary, it created tens of millions of new jobs: programmers, web designers, content creators, e-commerce, social media managers, data analysts, UX researchers — professions that didn't exist in 1995.
Even so, Sale didn't fully concede. He said: "Maybe I was wrong about the timing, but not about the direction." Inequality has risen. Some industries have died (print newspapers, travel agents, in-person bank tellers). Social media has created a mental-health crisis. And the environment... is still getting worse.
Today's lesson: AI vs jobs
In 2025–2026, a similar debate is erupting — only the protagonist now is AI.
The pessimists (the new Jeremy Rifkins — Geoffrey Hinton, Yuval Harari, and some Bloomberg economists who just sounded the alarm in May 2026): - AI will wipe out most office work within 5–10 years - Unlike the internet — the internet complemented humans, AI replaces them - "This time is really different"
The optimists (the new Kevin Kellys — Marc Andreessen, Reid Hoffman, Sam Altman): - Every technological revolution creates new jobs we couldn't have imagined - In 1995 no one predicted there would be a job called "prompt engineer" or "TikTok creator" - AI will amplify human capability, not replace it
Why this debate is hard
The truth is both sides are partly right:
Just as Sale predicted: many jobs will die. Bloomberg in May 2026 cited research — some businesses have officially replaced staff with AI (see the Klarna case cutting 700 customer-service reps, and content farms laying off writers en masse in 2024–2025).
Just as Kelly predicted: new jobs are appearing. In 2023 there was no "AI safety researcher" hiring on LinkedIn with a six-figure salary. By 2025 there were a few thousand.
But the key differences in 2026 vs 1995: - Speed: the internet took 20 years to penetrate the economy. AI is doing the same in 3 years. - Scope: the internet automated basic and connective work. AI automates thinking — the essence of knowledge work. - Learning curve: in the internet era, workers had 10–15 years to learn new skills. Now it's 1–2 years before the next model arrives.
The questions no one can answer for sure
Sale was wrong about the timing of collapse, but may be right about the direction. So with AI: - Are we at the start of a huge new wave of job creation, or at the start of the first time in the history of technology that destroys more jobs than it creates? - If new jobs are created, who will do them? Can a transport worker replaced by a self-driving vehicle learn to be a prompt engineer? - The question isn't "will AI take jobs" but "where does the income go when AI does the work?" — to the capital owners (who own the models + GPUs) or to the workers (with new skills)?
A personal lesson
If Kelly was right in 1995, then those who jumped in early to the internet got rich (Bezos, Brin, the early engineers). Those who stood warily on the sidelines got left behind.
In 2026, the same logic holds: instead of arguing "will AI kill jobs," the more valuable question is "Am I using AI to 10x my own productivity, or standing aside waiting to be 10x'd by someone else?"
Sale bet against the wave and lost. Not because he was stupid — he was smart and understood Luddite history better than anyone. But because when you bet against an accelerating technological wave, you're betting against the compound interest of innovation.
Betting against AI today is the same.
Sources
- Boing Boing (2021): "Kevin Kelly's 25-year bet with a neo-Luddite has been decided"
- Wired (March 1995): "Interview with the Luddite" — Kevin Kelly interviews Kirkpatrick Sale
- WSJ X (Nov 2025): "Pundits in the late 1990s offered all sorts of predictions about how the internet would affect jobs. For the most part, they were way off."
- Bloomberg (May 2026): "AI-Related Job Displacement Isn't Being Planned for Enough, Economists Say"
- Paul Krugman (1998, Red Herring): predicted the internet's economic impact would be no greater than the fax machine — wrong
- Jeremy Rifkin, The End of Work (1995) — predicted a "post-market era"