That's not what I said.
A nanomachine per se does not have to be an AI.
Any system, nano or macro, capable of creating a pipeline, will contain
within it many human-level intelligences. Any system, nano or macro,
capable of producing the variety of goods produced by the American
economy or the Japanese economy, will contain within it approximately
a hundred million human-level intelligences.
In a complex construction project, such as creating a pipeline, there are
many points at which a decision has to be made which requires human
judgment. In other words, whoever makes this decision has to
understand the whole project, and the purpose of his present task within
the project. He has to understand what he is doing in terms of a causal
model of the world, which is what I mean by human-level intelligence.
At present, such decisions are made by human beings. In the future,
they may be made by robots with AI. But the same decisions that
have to be made now will still have to be made in the future, and these
decisions will still require a mind with a causal model of the world.
In last night's post, about how to calibrate the idea of Extropian
Nanotech, I carefully avoided assuming anything about AI.
I assumed that all decisions requiring human judgment would be made
by humans, either in the design stage, or in the production stage.
However, this does not affect the point made above: the same decisions
that are made now will still have to be made in an economy based on
Extropian Nanotech, and those decisions will still require human
judgment.
Lyle