How Do You Even Tokenize Human Intelligence?
Human intelligence already works through input and output. Input is everything we read, see, hear, and feel. It is our senses gathering the world and our cognition organizing it into understanding. Before transformers ever existed, the human mind was already doing something very similar: taking in information, compressing it, relating it, and building meaning from it.
Output is what comes next. It is how we assemble all of that input into our best solution, our best advice, our best recommendation, our best direction, and our best effort. When a person responds with judgment, they are not producing random words. They are delivering a synthesized answer built from everything they have absorbed, processed, and understood over time.
In AI, tokens map to compute, and compute maps to energy. In humans, input and output map to time. Time spent reading, watching, listening, thinking, designing, deciding, and caring. That is why tokenizing human intelligence is not such a foreign concept. AI measures intelligence through compute and energy; humans express intelligence through cognition and time. The underlying mechanics are different, but the structure is not as different as people think.