Information Technology Reference
In-Depth Information
•
Machine mismatches can be detected through tautology. This is already done with
such ideas as a parity bit; a mechanism for assuring the correctness of a computer
word by indicating that there should be either an even or odd number of bits.
•
Programs on 'quantum' computers are bounded by operations that do not depend
upon knowing an interpretation. A 'quantum' computer is one that is based on the
idea of a 'qbit' or quantum bit. This works at the atomic level where 'superposi-
tion' is possible. Superposition means existing simultaneously in more than one
state (two in this case). A qbit has two simultaneous states and so every operation
is performed on all possible states simultaneously. This, in effect, allows massive
operational parallelism. Small-scale experiments have shown it can work in prin-
ciple. The problem, in practice, is the maintenance of the very special environment
in which it can happen.
•
Formal 'objects' (e.g. Windows in Object Oriented) will be stable but informal
'objects' (e.g. persons, chairs or games) will never be fully captured or be stable
because they are irrational sets (see example above).
•
It will not be possible to completely represent certain human functionality such
as natural language understanding on any machine that is not adaptable.
•
Increasing a training set for machine-learning algorithms will eventually cause
degradation in its recognition performance if the set includes irrational distinc-
tions. This is because there will arise increasing number of contradictions as
further training examples are given.
9.4
Inferential Semantics
From an engineering point of view the only information that can be experienced by
an individual is the result of the interaction of the individual's sense organs with
the world. This is not a passive view, since these organs are also controlled by an
inference engine; namely the human mind. It is only through inference and the senses
that we experience the world and relate to other people. So, like the computer, we
might be able to trace the sense of our understanding of the world through the tracing
of internal constructs to our senses. However, this would not be of any great help to
other people since it is unlikely that we are identical in the same way as two computers
that are constructed according to a defined engineering diagram are identical. If we
were to be different by as little as one bit we could not ever be sure that a 'program'
would mean the same if 'run' in different heads or that it would even 'run' at all. So
tracing and knowing the 'program' (or our internal constructions) is not very useful.
This puts some doubt on the value of tracing circuits in the human brain as a means
of understanding how 'it works'.
What could work, from a purely pragmatic point of view, is if individuals could
construct models of the world, and of other people, that were sufficient to meet the
needs of surviving in the world and with others. This model does not have to be
exact, just sufficient. However, to do this we have to extend our semantic model
to have another definition of meaning; a definition that does not depend upon the
