A friend of mine, sharp, senior, 25 years in consulting told me he can make an LLM write client deliverables in half the time. I asked him: what do you do with the other half? He did not have an answer. That silence is the story of the next decade.
In the late 1970s, calculators flooded the markets and researchers confirmed the fear: mental arithmetic declined. Parents and teachers were worried. Abacus classes started. Nobody missed long division. The people who won were not faster at math. They were the ones who knew what was worth calculating.
AI is echoing that pattern, but the stakes are different. AI is replacing the drafting, structuring, summarizing, pattern matching, creating text, images, videos, voice, that most knowledge work actually is. It is sitting in your chair and doing your work. The question to ask is how do you define your work? So what is still yours?
Let us talk about work. Literally. Newtonian mechanics.
In physics, work = Force × Displacement × cos θ. if you carry a heavy weight on your head and walk across a flat surface, the work you do in physics is zero. The force is vertical. The displacement is horizontal. The angle between them is 90 degrees. Cos 90° = 0. No work.
But you are exhausted. You are sweating. But on the axis that matters to the equation, you moved nothing.
This is exactly what AI is revealing about how many of us work. An LLM generates a full deck before the coffee gets cold. Enormous effort. Zero displacement on the axis that creates value. Meanwhile, someone else spends twenty minutes reframing a question. That moves things.
And no, you cannot prompt your way there. You can ask an LLM to think. That is the latest trend on Twitter. Socratic thinking, Feynman thinking. Do not tell the LLM what to do. Ask the LLM. Use first principles of thinking. Question it. Can you be my CTO, CFO, CMO?
The LLM will respond. You get your answers. Feynman's gift was not elegant answers. It was his refusal to accept the question as given.
That instinct, the irritation with unexamined premises, does not live in a model trained on consensus. The breakthroughs that reshaped our lives did not come only from processing more information faster. They came from someone willing to question what everyone else had agreed to stop questioning.
Original thinking is not a speed problem. AI solves speed. It does not touch the thing that matters, original thinking.
The calculator did not destroy mathematicians. The threat is not that AI will replace your job. It is that AI reveals how much of your job was never work, just weight carried sideways.
When I was a child, my mother told me: It is better to stand and drink water than run and drink milk. It’s a proverb in Telugu.
The question to ask is are you willing to stand and think and re-define how you work?