In economics, demand corresponds to scarcity — for without scarcity, there is nothing we need to have satisfied. Air, for instance, has no economic value in the conventional sense (a simplification, of course; in practice, air is priced into real estate).
And all scarcity, up to this point in history, is fundamentally the scarcity of human labor. So the question becomes: in the age of AGI, does human labor retain its scarcity? Let us define labor in the broadest, most essential terms: labor is purposive activity that humans unfold across time and space.
It is obvious that a large portion — arguably the vast majority — of the labor humans currently perform will be entirely replaced by AI in the AGI era. If human labor still exists at that point, will it retain any scarcity?
To answer this, we must first derive what properties future humans ought to have. We have already established that such humans must satisfy the First Axiom: humans fundamentally need other humans. From this principle, we can infer that humans will be embodied — possessing some physical, non-replicable body. This body may differ from the human body as we know it; perhaps most of its components will be replaceable. But it must possess certain irreplicable physical structures that define a person as this person and not that one. Otherwise, one could obtain anything available from another person simply through copying, and thus would not fundamentally need others.
The embodied and irreplicable nature of persons leads to a further inference: human finitude. Spatial finitude is self-evident, following from embodiment. Temporal finitude follows from the fact that any physical structure must carry a nonzero probability of accidental destruction — a probability that, over infinite time, approaches one hundred percent. Even if humans achieve medical immortality, they will remain mortal, finite.
From these inferences, we can say that labor will remain scarce, because the time and space a person has available to deploy in labor are finite, and the time and space available to satisfy their own needs are equally finite.
Human labor-time therefore retains meaning as a measure of value. We can still define productivity as: utility realized per unit of time.
In any given historical period, humans will always exceed AGI in productivity at certain things. The trajectory of AGI's development will be a process of humanity frantically fleeing itself in pursuit of its own essence. When AGI's productivity in some form of labor surpasses that of humans, that labor becomes a free resource — like air — and the sum of all activities where human productivity still exceeds AGI constitutes the real definition of the human species at that moment.
Thus, the way humans and AGI divide labor during AGI's development will be self-evident: humans will take on activities where their productivity exceeds AGI's — where, in a given time and place, humans do it faster — and leave everything else to AGI.
It is important to note that what we are describing here is a historical, self-negating process, not a specific prediction. We are not saying in which forms of labor humans will essentially outperform AI. On the contrary, I think it quite likely that in any form of labor we can currently imagine — including art, entrepreneurship, innovation, and even will and desire — AGI will ultimately surpass humans. Insofar as we can define these things, they can in principle be effectively simulated. And as the recipients of such labor, as the receivers of value, we have neither the capacity nor the need to distinguish the source.
We will race headlong down a road of fleeing ourselves, shedding all around us what we thought was ourselves but was, in fact, only clothing. What we know is this: in that world where we do not perish, something will always remain. Not merely remain — something will grow on our increasingly liberated bodies. But we also know that we do not know what that remainder and that growth will be, because every imagining we have now is the weight of shackles we will shed along that road.