I neglected at the time that coordinating information and putting it
into context requires processing power, and that the amount of
processing power required grows exponentially as the amount of
information increases. In other words, the more you know, the longer
you think about new facts. Any hypothetical intelligent computer
would be contrained by this limitation.
Intelligence is the ability to put things in context. Extrapolating
and predicting are but one facet of this. Extrapolation is the art of
putting the unknown in the context of the known.
The difference between an intelligent agent and a dumb recorder is the
ability to put new information in the context of existing knowledge.
It is in the nature of information that the more you have, the better
you are able to absorb new information. The information in a
technical paper or trade journal means nothing to a layman, but a
professional already has the knowledge needed to absorb it.
If you know more, then you require less packaging in the information
you recieve. A mathematician can see worlds in a simple equation, but
a normal person needs to read a whole book in order to understand the
information in that one equation.
In other words, the smarter a being is, the faster that being can
absorb new information. Thus, assuming a constant rate of data input,
the intelligence and information that a being posesses can grow
exponentially over time.
However, humans have inherent physical limits to the amount of
information that can be stored and retrieved. This works against the
pattern of exponential intelligence growth, with the exception of a
few unusual individuals.
However, computers do not have these limitations. Once computers gain
the ability to put information into context, there will be no stopping
their rate if intelligence growth.