Compu Think

“Will It Be Possible to Build a Computer That Can Think?
“Will It Be Possible to Build a Computer That Can Think?”

Peter Dey, 15 May 2003
The Information Technology revolution is occurring at an extraordinary

pace. What started 40 years ago with a computer the size of a room which
could perform simple arithmetic from punch-card inputs, has now evolved into
an essential business and entertainment tool. At this rate of evolution the
question has to be asked: Will it be possible for computers to think?
There are many definitions of the verb ‘to think’. With these definitions
come many questions and implications, such as “Does thinking imply
emotions?” Primitive animals can essentially only perform one task at a time,
and their set of possible tasks is usually quite limited, for example, hunting
and gathering. Given tens of thousands of years of evolution, modern society
is the end result. If computers were given this time, would they evolve to
become what society would consider to be sentient? Even at this point in
time, computers match, or even in some areas exceed, the benchmarks or
tests which have been created to assess the ability to ‘think’.
A definition of ‘think’ from the Merriam-Webster dictionary1 is "to

subject to the processes of logical thought". Every time an operation is
performed on a computer, it is run through the computer’s internal processes
(hence the “brain” of the computer being labelled as its processor). All
programs rely on logic to function – this is the sole purpose of a computer. If
one were to ask a computer what the answer to 1+1 was, and a result of ‘3’
were obtained, the computer would not be of any use. Society values
computers in society because of their ability to maintain logic in all situations.
However, does the ability to think imply the ability to evolve? Primitive
humans and many animals essentially have the ability to perform one task at
a time. They are also limited to a small set of tasks; for example, hunting and
gathering. In this way, they are “limited to their programming”. Instincts are to
mammals as programs are to computers; they’re simply a starting place. In
mammals, these instincts were allowed to develop, and modern society was
born after tens of thousands of years. Computers are only 40 years into their
evolution. Computers’ ability to adapt was limited by their programmers’
short-sightedness until the mid-1990’s.
Until the mid 70’s, computer programmers couldn’t grasp the idea of a
recursive subroutine (a section of a program which calls itself). Until this point
in time, subroutines were used simply as aides to the main program. It wasn’t
until Politzer Rozsa published "Recursive Functions in Computer Theory"2 that
programmers realised the potential they had previously been neglecting.
1
Merriam-Webster Online Dictionary, “think” – definition #9 (http://www.m-w.com/) 18/05/03
2
Rozsa Peter: Founder of Recursive Function Theory
(http://www.sdsc.edu/ScienceWomen/peter.html) 18/05/03 5:28 PM
In the mid 90’s, David B. Fogel began the development of a radically

new computer program – one that learned3. Dubbed ‘Blondie24’, the program
was given basic information: the size of a checkers board and legal moves;
but not which moves were good, or even how to win. Blondie had to deduce
this information on its own. Fogel played hundreds games with Blondie. With
each ‘generation’, Blondie became more intelligent as to which moves were
good and which moves weren’t. Blondie24 is now in the top 500 of checkers
players in the world4. This suggests that, with the right frame-of-mind a
programmer could write a program that could learn and adapt in any situation.
With the ability to learn and adapt would naturally come the ability to
make decisions5. If a computer program were taught what was ‘wrong’ and
‘right’ in a given situation, would it be able to move this information to another
similar situation? Does the ability to think imply the ability to make a
judgement? Blondie24 will put together pieces of moves it knows to be “good”
(i.e. have a heavier weighting) to counter a tactic it’s never seen before. In a
similar way, programs may soon be able to make judgements on a larger
scale.
In the 1980’s, Hans Moravec estimated that it would take 10,000 Cray-
2 supercomputers (the world’s most powerful supercomputer at the time) to
emulate the entire human brain6. A single Cray-2 is capable of 2 Giga-
FLOPS (Floating Point Operations Per Second). By this logic, 10,000 Cray-
2’s would yield 10 Tera-FLOPS. In 1998, IBM’s “Pacific Blue” was capable of
3.2 Tera-FLOPS7. The US-Backed ASC Initiative aims to build a 100 Tera-
FLOPS supercomputer by the year 2004. Given a mastermind such as David
Fogel, the emulation of the entire human brain could well become a reality.
While hosting a discussion panel, an individual suggested that the

ability to think implied the ability to be bored (from lack of material to think
about). The example was given:
“If you lock a human in a room for 48 hours, the human will get bored
and irritated. If you leave a computer unattended, it still responds to
your commands after 48 hours.”
Currently, when a computer’s resources are unused, it executes an

‘idle loop’ program. This ‘idle loop’ means that the computer’s processor is
running at 100%, all the time8. Essentially, a computer doesn’t have the time
3
Blondie24, Playing at the Edge of AI (http://www.comp.hkbu.edu.hk/~cib/2002/cib_iss1f.pdf),
16/05/03 3:24 PM
4
Evolving Solutions that are Competitive with Humans, David B. Fogel, pg. 3
(http://www.isd.mel.nist.gov/research_areas/research_engineering/Performance_Metrics/Per
MIS_2002_Proceedings/Fogel.pdf) 18/05/03 3:27 PM
5
Asimov’s Laws of Robotics, Roger Clarke
(http://www.anu.edu.au/people/Roger.Clarke/SOS/Asimov.html) 18/05/03 6:07 PM
6
Taking Sides: Clashing Views on Controversial Issues in Science, Technology & Society, 5th
Ed, pg 252, Thomas A Easton, McGraw-Hill
7
Computer History (http://home.earthlink.net/~mrob/pub/ch-machines.html) 18/05/03 3:57 PM
8
Microsoft Windows 2000 SP3 5.00.2195, Task Manager, ‘Processes’ tab, 18/05/03
to get bored. If this idle loop did not exist, the operating system may go
haphazard, in a similar way to bored human teenagers.
Silicon technology has its limits. Chips can only go so fast, and can
only get so small – you can only get down to the size of an electron. The next
stage for computing is biological, or neuron processors9. Neurons are the
brain’s “circuitry”. They have many things in common with silicon chips – both
transmit electrical signals. Seymour Cray himself, founder of Cray Computer
and creator of the Cray Supercomputer, said that the next step for computing
was to use DNA and proteins, “just as Nature does”10.
However, there are problems with using neurons to transmit signals:

while silicon chips perform predictably, neurons can behave chaotically.
Perhaps this inherent chaos in the human brain is what allows us to be
creative and think outside the square. Soon, using neuron technology, even
computers may have this ability.
Computers currently have many abilities that are regarded as thinking;

by the general public and many dictionary definitions of the word. However,
some still say that computers will never fully be able to emulate a human's
intelligence. With the advances that have been made over the last 40 years
of the evolution of computers, it is not safe to assume so. Advances such as
'Blondie24', a program that learns, and now the move into quantum and
biological computing, indicate that the ability of a computer that can think is in
the not-to-distant future.
9
“A New Breed of Thinking Computer?”, BusinessWeek Online
(http://www.businessweek.com/1999/99_25/b3634137.htm) 18/05/03, 4:09 PM
10
Supercomputers, Don Calle (http://ei.cs.vt.edu/~history/SUPERCOM.Calle.HTML) 18/05/03

Compu Think

Uploaded by

Copyright:

Available Formats

Compu Think

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Compu Think

Uploaded by

Copyright:

Available Formats

“Will It Be Possible to Build a Computer That Can Think?

“Will It Be Possible to Build a Computer That Can Think?”

The Information Technology revolution is occurring at an extraordinary

A definition of ‘think’ from the Merriam-Webster dictionary1 is "to

In the mid 90’s, David B. Fogel began the development of a radically

While hosting a discussion panel, an individual suggested that the

Currently, when a computer’s resources are unused, it executes an

However, there are problems with using neurons to transmit signals:

Computers currently have many abilities that are regarded as thinking;

You might also like