Download How useful is an intelligent computer

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Human–computer interaction wikipedia , lookup

Computer Go wikipedia , lookup

Visual Turing Test wikipedia , lookup

Technological singularity wikipedia , lookup

AI winter wikipedia , lookup

Artificial intelligence in video games wikipedia , lookup

Wizard of Oz experiment wikipedia , lookup

Embodied cognitive science wikipedia , lookup

Chinese room wikipedia , lookup

Turing test wikipedia , lookup

Intelligence explosion wikipedia , lookup

Existential risk from artificial general intelligence wikipedia , lookup

Ethics of artificial intelligence wikipedia , lookup

History of artificial intelligence wikipedia , lookup

Philosophy of artificial intelligence wikipedia , lookup

Transcript
How useful is an intelligent computer ?
Rieks op den Akker
Twente University, Enschede 7500AE, the Netherlands,
[email protected]
Although the notion of machine intelligence (or the synonymous artificial
intelligence) has been an issue of heavy debating for some decades (’is it possible
to make intelligent machines?’), it seems that there has always been agreement
among the debaters that the abilities to have a conversation in natural language
-that is: understanding natural language and being able to generate a coherent
text- are among the most typical phenomena of human intelligence. Indeed,
human language is the manifestation of intelligence and when other primates
were as intelligent as human beings they would have invented and developed a
language like we have.
”As I can see it, any ordinary person who can understand an ordinary
conversation must have already in his head most of the mental power
that our greatest thinkers have.”
A good composer, for example, has to master many skills of phrase and
theme- but those abilities are shared, to some degree, by everyone who
talks coherently. An artist also has to master larger forms of form - but
such skills, too, are shared by everyone who knows good ways to ”tell a
tale”.
These two quotations are from ”Why People Think Computers Can’t” an
essay in which Marvin Minsky, one of the leading thinkers that have shaped the
field of Artificial Intelligence, tries to unravel the minds of those peoples who
don’t believe that machines will ever ”really think” ([4]). The people who built
the first computers were engineers concerned with huge numerical computations.
’That’s why these things were called computers’. Alan M. Turing (1912-1954)
was one of those scarce scientists at the fringe of people who envisioned what is
now called AI ’because they realized that computers could manipulate not only
numbers but also symbols.’ 1
Turing proposed to set up a test to settle the question whether a machine
could be considered intelligent: the Imitation Game. An interrogator may ask
questions to two creatures; one of these is a machine, the other one is a human
being. Based on the answers the interrogator should try to decide who is who.
When the machine is able to convince the questioner that he is the human
being he should be called intelligent (see Turing’s Computing Machinery and
Intelligence, [6], that also appeared in [1] and in [2].) Most critics agree that the
1
On Minsky’s web site http://web.media.mit.edu/ minsky/ you’ll find the complete
text of this paper and many more publications of the author of The Society of Mind
([5]).
Imitation Game is a good test for machine intelligence: the ability to understand
questions and answer them correctly is a strong sign of intelligence. In 1950
Turing had rather high expectations about the progress made in the field of
artificial intelligence.
I believe that in about fifty years’ time it will be possible to programme
computers, with a storage capacity of about 109 , to make them play
the Imitation Game so well that an average interrogator will not have
more than 70 per cent chance of making the right identification after five
minutes of questioning.
At the time of this writing, more than fifty years after the publication of the
imitation game, no computer system has passed Turing’s test. For some critics
of AI, like D.C. Dennett this means that the question is still open - and it remains open until a computer will pass the test - others like J.R. Searle and also
S. Stenlund do not consider Turing’s test a valid test for the existence of machines that are able to think like people do and try to answer the question in a
more principal, philosophical way by means of an analysis of the basic concepts
involved: mind, machine, and meaning. (See the contributions of Dennet and
Searle to the special issue of ’Wijsgerig Perspectief ’ [9]), and Searle’s ”Minds,
Brains and Programs” ([7, 2]).
The development of technology has always had a strong impact on the way
man conceives himself. In the age of ICT man sees him self as an information
processing device. In the age of interacting computers man conceives man as a
conversational agent performing speech acts. The Swedish mathematician and
Philosopher Sören Stenlund noticed the strong relation between what might be
called ’the machine conception of the human mind’, and ’the calculus conception
of human language’. (see his essay ’Language, Mind and Machines’ in [8]). In
this ’calculus conception of language’, the use of language is conceived as ’symbol manipulation’ (see the reference to Minsky above) and as something distinct
from the ’manipulation of numbers’. No doubt, the relation between man and
his environment and between man and him self and other people change through
the development of technology.
Ambient intelligent environments imply a completely new dimension in the
relation between man and technology. Technology takes - among others - the form
of a Personal Assistant. Man do not use the computer, they do not program it,
they have a personal relation with it. But how can such a relation be established ?
What is the common basis on which such a relation can be built? The statement I
want to make is this: will there be a relationship between man an his environment
that could be called ”personal” then the environment should be able to learn
how to communicate with its inhabitants by means of natural language. This
language is not a built in language, predefined by the makers, but it is shaped
through the interaction between the inhabitants and the environment. A number
of questions concerning human conversations and the way personal relationships
are established by means of conversations have to be answered in order to make
such personal environments become reality. A few of these questions are:
– How does a Personal Assistant learn the meaning of new words and new
language constructs?
– How does an environment learn when he is addressed by an inhabitant and
how he has to address the inhabitants in an acceptable way?
– How does a Personal Assistent learn when he is allowed and expected to take
initiative to act on his own and when he has to ask his ’master’ permission
to act following his own insight in the situation ?
In general, a personal relation can only be established when the environment,
or assistant is cognitively and linguistically flexible enough to be able to learn
new language and new behavior. In other words when the relation invents new
relations between signs and the things they signify: things, events, actions, feelings, based on common experiences. The idea of ambient intelligent environments
and personal assistants implies a challenge for the development of techniques for
learning social conversational behavior.
Now, I come to the final issue I want to raise -whence the title of this contribution. It concerns the distinction between the personal and the technical realm.
Technical things are instances of a general abstract idea. The technical idea completely specifies the instances, and they are all exactly the same. That is what
makes a technical device something that we can count on and a device that we
can use for our aims. The technological view on language conceives a sentence
as some abstract thing, as something that has a meaning outside its situation
of use and whose instances we can use and re-use in ’appropriate’ situations. A
person, on the contrary is something unique: an individual being that is not an
implementation of some idea, but that itself defines what it is and becomes. The
very idea of a Personal Computer has always been an intriguing one. Suppose
that we were able to make such an intelligent being, some being that becomes
our personal assistant. Will we still consider it as a useful thing? Or will it fall
outside the scope of things that we call technical devices; a means that we use
for achieving something, some end that is outside the value of the relation itself.
Can we use our Butler, the way we use a technical device, and have a personal
relation with it? Turing’s beliefs about the possible capabilities of machines were
not limited to such activities as playing the imitation game. He considered it very
well possible to develop a machine that would relish the taste of strawberries and
cream. However, Turing also thought it would be ’idiotic’ to attempt to make
one (see [3]).
Why should we bother ourselves creating a personal assistant? Just, because it’s
fun, or also because it’s really useful?
References
1. A.R. Anderson (ed.) Minds and Machines. Contemporary Perspectives in Philosophy
Series. Prentice-Hall, Inc, Englewood Cliffs, New Jersey, 1964.
2. R. Cummins and D.D. Cummins (ed.). Minds, Brains and Computers; the foundations of cognitive science; an anthology. Blackwell Publishers, 2000. (this reader
contains, a reprint of Searle’s ’Minds, Brains and Programs’ as well as Turings
’Computing Machinery and Intelligence’).
3. K. Gunderson. The Imitation Game In: Minds and Machines. Contemporary Perspectives in Philosophy Series. Prentice-Hall, Inc, Englewood Cliffs, New Jersey,
1964.
4. M. Minsky. Why People Think Computers Can’t. In: The AI Magazine; Fall 1982:315.
5. M. Minsky. The Society of Mind. Simon and Schuster, New York; 1985.
6. A.M. Turing. Computing Machinery and Intelligence. In: Mind, LIX, 1950.
7. J.R. Searle. Minds, Brains and Programs. In: The Behavioral and Brain Sciences,
3, 1980.)
8. S. Stenlund. Language and Philosophical Problems, Routledge, London and New
York, 1990.
9. Menselijk denken en computers; Wijsgerig Perspectief op maatschappij en wetenschap; 24(4), 1983.