|
Summary of The Inconceivable Thoughts
We know what we are – but not what we may become (Shakespeare)
In Ray Kurzweil’s book The Age of the Spiritual Machines this
passage appears as an introduction to a future society in the year
2099. In this society, man- and machine intelligence cannot be distinguished.
The possibility of artificial intelligence (AI) is seen as a matter
of having sufficient computational power. Kurzweil’s thoughts
seem to comprehend intelligence and consciousness in an unduly unproblematic
way. As many scientists share his views on AI, I find it necessary
to investigate whether AI is as unproblematic as it seems to be.
The basis for these investigations is the notion of intelligence
used by Howard Gardner in his theory of Multiple Intelligences (MI).
My hope is that this psychological notion of intelligence will display
a different picture on consciousness and as a result of AI.
Artificial intelligence can be divided into two paradigms - classical
AI and connectionism. These paradigms have differing opinions on
the perspectives of AI. While classical AI claims to synthesize
the human consciousness, connectionism takes a more subtle approach,
only claiming a possibility of simulation. Classical AI perceives
intelligence as a matter of symbolic representation. Symbolic representation
exhibits numerous problems that can be seen as results of the attempt
to mechanize reason. The Paradox of Mechanical Reason emphasizes
that …if a process or system is mechanical, it can’t
reason; if it reasons, it can’t be mechanical . Furthermore,
an explanation of consciousness by means of symbolic representation
leads to the problematic notion that once syntax is established,
semantics will emerge. On the other hand, connectionism does not
see representations as a direct symbolic unit, but instead as distributed
through the interaction of neurons. This complicated pattern of
conceptual representation and their relations forces us to a description
of understanding on a subconceptual level, and therefore it does
not seem sufficient as an explanation of consciousness. Both paradigms
appear to rely on some external criteria for intelligence, and in
the end, this results in the reduction of intelligence as a matter
of behavioral patterns. The question that comes into mind is whether
this psychology of the mind, represented by any of the two paradigms,
seems adequate when the psychological conception of consciousness
as exemplified by Howard Gardner is included. In contrast to the
hitherto dominant paper and pencil tests, MI seeks to establish
a biological foundation for the notion of intelligence. Gardner
sees the brain as a biopsychological unit that processes information.
Gardner exhibits eight intelligences that he imagines belong to
specific neural parts of the brain. At the same time, he seeks to
come to terms with the notion of intelligence as a construct. The
notions of intelligence thus give the impression to be blurred.
Furthermore, MI presents a hazy notion of the way that meaning comes
off. On one hand, meaning occurs as a result of simple manipulation
with symbols related to the computational capacities of the mind
- on the other hand, Gardner stresses that his theory avoids the
philosophical epistemic problems. As we have seen, the notion of
symbol manipulation does not conceal these problems. The possibility
of AI is rejected on an explanation exhibiting the inconsistency
of MI.
On the assumption that the notion of computations is a feature
of both AI and MI, a discussion of David Chalmer’s (DC) notion
of consciousness appears to transpire naturally. With a notion of
the mental properties as organizational invariants (OI), the hard
problems of consciousness are dealt with. Given the properties of
OI, we can describe the structure of the brain and due to the principle
of structural coherence, we can describe the consciousness. In all,
consciousness is only a matter of a functional organization and
is not dependent upon the substance in which this organization is
implemented. DC imagines that the principles are consolidated into
a single theory of information. As such, conscious properties and
material properties can be seen as aspects of information. Chalmers’
theory brings up a number of matters. In relation to the scope of
this paper, I find the discussion on implementation most relevant.
As an opponent to DC, John Searle (JS) stresses that the substance
exposes an importance in the matter of consciousness. The Chinese
Room argument serves to underline the idea that semantic does not
take care of itself just by implementing a formal rule. Observed
from the outside, the answers from within the room give the impression
that the room understands Chinese even though this is impossible
according to JS. JS does not reject that the brain possesses computational
properties but simply emphasizes that this does not sufficiently
explain how the brain comes to produce mental states. As an answer,
DC argues that programmes can be seen as syntactic objects, whereas
implementations cannot. On the contrary, JS argues that an implementation
must possess the same causal abilities as the brain but he does
not give any criteria of the way this is to come about. According
to JS, the explanation is to be found in the fact that consciousness
and mental states are biological phenomena. Mental states are both
seen as caused by the neurophysiological processes and as features
of the brain. These conceptions reveal a causal interaction and
at the same time an identity relation between the mental properties
and the brain. However, JS states that the mental states supervene
on the material states. While at the same time being seen as an
identity relation, the idea of this is obscure. The problem of causality
between the mental states and the material states seems to lead
either to a kind of dualism and as a result of this to an incomprehensible
notion of causality or if rejecting this causal relation, to a form
of identity thesis with attendant epiphenomenalism . Both DC and
JS can be characterized as nonreductive physicalists. As such, both
theories and as we have seen the AI and MI are faced with the problem
of avoiding to reduce the mental states to the material states while
simultaneously explaining the psycho-physical relation in a comprehensible
manner. Thomas Nagel (TN) beholds the psycho-physical relation in
a quite different way. Subjectivity and objectivity are seen as
diametrical perspectives. The subjective aspect of the mental states
is only comprehensible from the creature itself. Thus, What it is
like to be a bat cannot be comprehended from a human point of view.
Science aims at giving a description of reality as an objective
reality. This objective ideal is unification in a point of view
from nowhere. The essential subjective character, however, is the
specific point of view. As a consequence, we are faced with a need
for a new understanding of reality that illuminates instead of eliminates
the subjective point of view. Though the objective phenomenology
of TN in many ways is problematic, I find his considerations on
the objective ideal rewarding in this discussion on consciousness.
With this paper, my initiate aim was to enlighten the discussion
of AI through the findings on consciousness in MI. Sad to say, MI
does not bring anything new to the discussion. In trying to explain
the impossibility of AI, It rather seems that the theory is stretched
in such a way that it provides more questions than answers. My investigations
have not been in vain. As we have seen, the question of AI embarks
on a comprehensible notion on consciousness. As it has turned out,
the initiate quote of Shakespeare is by no means obvious. A new
perspective is called for. These considerations stress the need
for collaboration between different disciplines in a search for
an answer. In humble appreciation of the powers of the mind, we
must face these future examinations on consciousness.
|
Send siden til en ven!