Marshall McLuhan had one idea that outlived him. Every technology is an extension of the body. The wheel extends the foot. The book extends the eye. Clothing, the skin. Electric circuitry, the nervous system.
He had a second idea that gets quoted less. Every extension is also an amputation. The wheel carried the foot further. We walked less. The extension serves the faculty. It also puts it to sleep. McLuhan wrote this before the computer reached an office. Everything that followed obeyed both laws.
Four waves
The computer and its software came in four waves. First the computer itself, a box on a desk. Then the internet, which joined the boxes. Then ERP, which put the company inside one system. Then SaaS, which moved the system to the cloud and charged by the month. Each time we complained for a quarter. Then we could not imagine work without it.
Each one extended something. The computer, memory and arithmetic. The internet, reach. ERP, discipline. I am still not sure it extended anything else. SaaS, distribution. The human’s ability to execute improved. Execution was never the upper hand. Thinking was.
Memory and arithmetic look like thinking. They are not. They are the head’s hands. Every tool so far extended the hands. None touched the thinking. Each one also took something. No spreadsheet ever asked why. No ERP did. They waited. We decided. They executed. We were the only ones in the room who could ask.
The tools sorted us
What did the humans do? The curious took the hours the tools gave back and went deeper. We have a word for it. Jigyasa. The desire to know. At work, it is the person who reads the footnote nobody asked them to read. The tools sorted us. Not by skill. By appetite.
The rest settled. The tool did the work. The comfort zone appeared. It was warm. Nobody was thrown out of it till the companies decided to. The appetite decided.
The tool that asks back
Now it is not the same. AI does not only execute. It does the first draft and the repetitive job. Then it does something no ERP ever did. It asks a question back. Ask it for a proposal. It asks who will read it. No system ever did.
Where does its question come from? From every word ever written down. Its question is the average of our questions. It is never the next question. The next question has not been asked yet. So it is not in the data.
Someone will say our next questions are also built on old ones. True. But there are two kinds of question. One asks because it does not know. Who will read this? When is it due? The machine can ask those. They are retrieval with a question mark.
The other kind asks because the answer is not enough. Why this? What if not? The question nobody has asked yet. That kind fuels thinking. It comes from someone who stands to lose if it goes unasked. The machine stands to lose nothing. It cannot ask the second kind. It can only produce the shape of it.
The confusion. The hype is beyond hyperbole. It’s time we find a new word to describe it. Hopefully the investor community will have a new word soon. The confusion is because we measure one object on two scales. On execution, AI wins. On thinking, it loses. One day it is a co-worker, the next a philosopher, the next a parrot.
The machine’s question is our own question, averaged and reflected back. We see a face asking. We do not recognize it. We call it a mind. It is a mirror.
What the mosquito knows
Can it work without data? No. It has read everything. It has lived nothing.
A newborn knows its mother. Within days. Before it has a word. What data did it have? The mother arrived. Something inside already knew. A biological neuron. Evolved over millions of years.
The mosquito has been here for a hundred million years. What it needed to know was built in. No prompt. No dataset. No update.
They were trained too. Both of them. The training data was every ancestor that died. Evolution is a dataset whose cost is extinction. AI is a dataset whose cost is electricity. A mathematical neuron. Nothing died for it.
Hans Moravec saw this in the eighties. What is hard for us is easy for the machine. Chess. Calculus. What is easy for us is hard for it. Recognizing a face (beyond cats and dogs). Walking across a room. Picking up a cup. His explanation was evolution. The easy things were paid for over millions of years and run below thought. The hard things are recent, thin, and simple to write down. So the machine took the recent things first.
Knowledge has two sources. What we were told. What we were built with. The machine has only the first. The mosquito has only the second. Humans have both.
It will sort us
So will AI conquer? Conquer means to take what was held. We held the question. AI will not take it. It will sort us. This time sharper. With ERP, you stopped doing. With AI, you stop asking. The seat where thinking happened will still be occupied. Just not by you.
Nobody will notice. The loss does not feel like a loss. V. S. Ramachandran, the neuroscientist who mapped how the brain handles a lost limb, showed why. His amputees still felt the missing hand. Some felt it clenched, in pain. A phantom. The person who has stopped asking will feel the same. Not an absence. A phantom. They will say, I thought about it. They will mean, I read what it said.
Ramachandran eased the pain with a mirror. The patient saw the good hand where the lost one should be. The pain stopped. Now ask what the pain is in our story.
The machine always answers. A human can hold a question open. Sit in it for a week. Say, I do not know yet. That discomfort is where every next question comes from. Jigyasa is the willingness to stay in the question and the problem.
Ramachandran showed the real mirror. The hand in it was the patient’s own. We are running to AI for answers. It comforts you instantly. You do not know whether the answer is relevant or right. It takes away the discomfort of staying without one. Is AI the real mirror?
Every extension is an amputation. AI is the first extension of the asking. By McLuhan’s own law, it will numb what it extends.
By using AI, are you amputating your thinking?