Quote Collection · 9 quotes · the augmentation-vs-replacement debate in the words of the people building and studying it. Paired guide: How to Keep Judgment When the Model Is Confident.
Why this collection
No question hangs over this decade like this one: does capable AI replace human work or amplify it? The people closest to the technology are split — and the split is instructive. The optimists frame it as augmentation; the realists warn about the pace of disruption; the cautious warn about trusting it too much. Read together, these lines aren't a consensus — they're a map of the live debate, from the researchers who built the field to the operators deploying it. Each is verified for wording and attribution; where a famous line has contested authorship, we say so rather than guess.
The quotes
"AI is the new electricity." — Andrew Ng, computer scientist, co-founder of Coursera and Google Brain (Stanford GSB, 2017)
"The future is already here — it's just not evenly distributed." — William Gibson, novelist (widely attributed; on how new technology spreads unevenly)
"AI is more profound than electricity or fire." — Sundar Pichai, CEO of Alphabet (2018)
"Rather than artificial intelligence, I think we'll augment our intelligence." — Ginni Rometty, former CEO of IBM
"AI is not that big, scary thing in the future — it's here with us." — Fei-Fei Li, computer scientist, co-director of the Stanford Institute for Human-Centered AI
"AI will increasingly replace repetitive jobs — blue-collar and white-collar." — Kai-Fu Lee, AI researcher and investor, author of AI Superpowers
"The future depends on some graduate student who is deeply suspicious of everything I have said." — Geoffrey Hinton, deep-learning pioneer
"AI will be either the best or the worst thing to happen to humanity." — Stephen Hawking, physicist (Cambridge, 2016)
"AI is a fundamental risk to the existence of human civilization." — Elon Musk, entrepreneur (2017)
A note on one line you'll see everywhere
The popular adage "AI won't replace you — someone using AI will" circulates constantly and is worth knowing, but its authorship is genuinely contested (it's been attributed to several people and echoes a point made by economist Richard Baldwin). We include it here as a widely-repeated saying of uncertain origin rather than pin it to a name we can't confirm. That honesty is the point: in a field moving this fast, a confidently-attributed quote is often the least trustworthy kind.
How to read this set
Notice the axis. Ng, Pichai, Rometty, and Li sit on the augmentation side — AI as a general-purpose amplifier of human capability. Lee sits on the disruption side — real replacement of repetitive work, fast. Hawking and Musk sit on the risk side — enormous upside or enormous downside. Hinton's line is the quiet meta-point that outruns all of them: the people who built this are the first to say they might be wrong, and the future belongs to whoever questions them. The useful stance isn't to pick a camp; it's to hold the augmentation opportunity and the disruption/risk warnings at once — which is exactly the Stockdale-Paradox discipline applied to technology.
Related mental models
- The Bitter Lesson — why the capability keeps compounding (so "AI is the new electricity" ages well).
- Jagged Frontier — why "replace vs. augment" has no single answer; it depends, unevenly, on the task.
- Amara's Law — we overestimate AI short-term and underestimate it long-term; half these quotes are arguing across that gap.
- Stockdale Paradox — hold the opportunity and the brutal facts simultaneously; the right posture toward this whole debate.
- Second-Order Thinking — the disruption quotes are really about the consequences of the consequences.
FAQ
Are these quotes verified? Yes — each was checked for wording and attribution, and the one line with contested authorship is flagged as such rather than misattributed. Why include both optimists and pessimists? Because the collection's value is the debate, not a verdict. The honest picture is that the field's own leaders disagree. Which guide pairs with this? How to Keep Judgment When the Model Is Confident — the practical companion to the "trust it too much" worry underneath several of these lines.