Download Is the search for computer-based artificial intelligence an

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

Turing test wikipedia , lookup

Kevin Warwick wikipedia , lookup

Artificial intelligence in video games wikipedia , lookup

Agent (The Matrix) wikipedia , lookup

Adaptive collaborative control wikipedia , lookup

Human–computer interaction wikipedia , lookup

AI winter wikipedia , lookup

Technological singularity wikipedia , lookup

Embodied cognitive science wikipedia , lookup

Intelligence explosion wikipedia , lookup

History of artificial intelligence wikipedia , lookup

Philosophy of artificial intelligence wikipedia , lookup

Existential risk from artificial general intelligence wikipedia , lookup

Ethics of artificial intelligence wikipedia , lookup

Transcript
Is the search for computer-based artificial intelligence an
appropriate pursuit for science; that is, is it in the best interest of
mankind to build a human machine?
Hector Alfaro
UFID 6667-9931
February 16, 2006
CGS 3065: Legal and Social Issues in Computing
Section 8213
Alfaro 1
Thesis
Humans are both fascinated and terrified of a computer having means by
which to self evolve. A vast number of books, movies, plays, talks and classes have
been created in light of Artificial Intelligence (AI). But along with the evolution of
this technology in recent decades, concerns have risen. Specifically: is it possible
to build a computer able to, not only think, but also feel? Would such a computer
gain control over the human race? If and when these machines are popular, how
will one properly differentiate between a machine and a human? Fear of the
unknown has caused more and more individuals to question the need for
research in the field of Artificial Intelligence. Rather than building a machine that
could destroy mankind, why not evade the idea altogether?
The basis for the many concerns that surround Artificial Intelligence is
ignorance. Humans are simply afraid of what they do not know. More
importantly, however, they are captivated by it. A large number of scientists have
recently turned their attention to Artificial Intelligence in hopes of clearing doubt
and instructing the masses on the possibilities of a hyper intelligent, self-evolving
machine (Drexler 75). Advances in Artificial Intelligence would benefit humanity
immensely, from applications in everyday mathematical computations to
advances in the medical field that could save lives (Lok). It is the goal of this
paper to instruct citizens of society on the current advances of Artificial
Intelligence and present future applications for the technology. In so doing, this
paper hopes to clear the many doubts that the average person has on the over
reliance
of
computers,
infrastructure.
specifically
computers
that
change
their
own
Alfaro 2
Introduction
Recent debate suggests that it is not in the best interest of mankind to build a
human–rather, humanlike–machine. In the film Bicentennial Man, the story of
Andrew Martin is told. At the beginning of the film, Andrew is a simple
household robot assigned tasks that all other robots can do: chores, caring of
children, repairing of broken objects. The difference with Andrew is that he
eventually becomes unlike the other robots of his kind. He develops skills that
require human mental (and emotional) capacities. Andrew learns to feel and, as a
result of his emotions, makes works of art. Andrew outlives the Martin family
who bought him, but he becomes so humanlike that he starts petitioning the
government to give him human rights. Andrew wants to be able to own property,
make money from his work and, more importantly, wed.
The main plot of this film is based around Andrew’s battle with the legal
system to acquire rights that are reserved for humans. The bureaucrats in
Bicentennial Man immediately deny Andrew’s requests. They ask why a robot
should be allowed human rights, no matter how humanlike it is. They believe that
despite Andrew’s clear ability to think and feel, he is still mostly metal, built in a
factory, and his abilities are just a good imitation of human behavior.
Andrew returns to the same board of government officials time and again
throughout his two-hundred year struggle to gain human rights. He reconstructs
his body, substituting all metal parts with flesh-like ones to the point that the
majority of his mechanisms are biological. Andrew wonders about robotic parts
in humans and wonders why they are treated differently than him.
Alfaro 3
Bicentennial Man is complete fiction, but it accurately portrays discussion
that lies ahead regarding Artificial Intelligence, as well as reasonable arguments
for both sides of the issue. Like in most films, Bicentennial Man has a bittersweet
ending in which Andrew is classified as human and given the rights that all
humans have; most importantly, the right to love. Present day society is a long
way from accepting artificial beings as human. This paper will present research in
the fields of Artificial Intelligence and sociology, as well as discuss fictional works
that give insight into a time not so far ahead. While the research presented in this
paper will not portray fully intelligent agents able to pass the modern-day Turing
test, it will present research that is well on its way to achieve it. This paper
intends to accomplish its goals by presenting the reader with enough information
to decide if the search for computer-based artificial intelligence is an appropriate
pursuit for science.
Discussion of Research
Debates in the field of Artificial Intelligence did not emerge recently. When
the Turing Test was introduced in 1950, there were many objections including
one by Professor Jefferson Lister: "not until a machine can write a sonnet or
compose a concerto because of thoughts and emotions felt, and not by the chance
fall of symbols, could we agree that machine equals brain" (qtd. in “Turing Test”).
Many individuals in modern day society share Professor Lister’s thoughts
regarding artificial agents. To perform tasks that require emotion is to say that
these tasks and expressions require a human. An artificial agent cannot simply
duplicate the complex emotional machinery of the human soul. A slightly
Alfaro 4
different view with the same goal suggests that the mind and the brain are quite
different things. Sociologists often compare humans to animals in an effort to
correlate and explain behaviors. Animals and humans both have brains capable
of executing tasks and, for some animals, even retaining adequate amounts
information; but only humans have what sociologists call a social mind (Wolfe
1076). It is the mind – not the brain – that can create art that cannot be
duplicated by a series of calculations. In that sense, some claim, artificial agents
could never be regarded as humans. Computers are thought of as giant brains,
capable of performing these calculations quickly and accurately. The emergence
of Artificial Intelligence, however, suggests that computers able to change
themselves tend to exhibit more mind-like behavior; by sociological definitions of
human abilities, these agents are exhibiting humanlike behavior.
Presently, working with an Artificial Intelligence agent is very much like
teaching a child. Like children, these agents are intended to grow and produce
their own thoughts in the future. The fear with this issue is that having
experienced the rapid advances in technology, an artificial agent’s mind will
similarly increase in complexity and intelligence. A hyper intelligent agent would
gain control of its thoughts. Multiple agents could then share their abilities.
When a machine gains awareness of its abilities in comparison to a human’s
mental ability, what will it do? Will it realize that “two heads are better than one”
and decide to team with other artificial beings? The question of how to control
beings that can outthink any human arises. Not having gathered enough
information in the field results in there being simply no answer. Those against
the advancement of Artificial Intelligence believe that, from an evolutionary
Alfaro 5
standpoint, machines can overthrow the human population (Drexler 173). Isn’t
this enough reason to stop research in artificial intelligence?
Some individuals do not share the fear that robots may overthrow the human
race; rather they fear interaction with a robot (Breazeal 3). Perhaps the reason for
this fear is that current research only explores so much. The most advanced
agents to date do not have the ability to pass the Turing Test, which states that:
After a prolonged keyboard ‘conversation’ with [an artificially intelligent
agent]—perhaps touching on literature, art, the weather, and how a mouth
tastes in the morning—it might that you could not tell which was the person
and which the machine. If a machine could converse this well on a regular
basis, then Turing suggests that we should consider it genuinely intelligent.
Further, we would have to acknowledge that it knew a great deal about human
beings (Drexler 72).
In a conference on artificial agents, Brezeal explains that conversation with
Kismet, one of the most advanced and personable intelligent agents today, is
composed of some awkward pauses and interruptions. Kismet does not speak
English. Its interaction with humans is based on the tonality of their voice and
body language—Kismet also responds in this manner. This sort of interaction
does not come completely natural to humans. While a human does understand
the importance of tonality, he or she will rely on the words being spoken over
body language. It is unnatural for someone to rely solely on tonality and body
language. Humans are accustomed to verbal interactions with one another and
adjusting one’s behavior for an agent seems alien.
Despite all the negative what-ifs that surround Artificial Intelligence, the
benefits that artificial agents will bring to society vastly outweigh the negative
aspects and worries associated with them. Advances in the field of Artificial
Intelligence and bionics have already saved human lives. Everyday people that
Alfaro 6
have suffered life-threatening injuries have artificial limbs or body parts due to
research in this field (Perkowitz 86). In a future society, saving humans from
drowning, minimizing automobile accidents, and assisting the elderly will all be
the job of the artificial agents it may house. Machines capable of rationalizing
solutions to situations due to its environment would be able to react more quickly
and efficiently than any human ever could.
At the University of Florida alone, there are three very important research
projects being conducted that have very useful applications in a not-so-distant
future society. Dr. Lok leads a group that has created intelligent agents made
possible through the use of virtual reality. These agents teach medical students
how to properly communicate with human patients. Whereas in the past medical
students had to wait until their residency or years of medical school to practice
scenarios in which they were evaluating a patient, now it is possible in the early
years of their career, thanks to Virtual Patients. This project intends not to avoid
human interaction, as some individuals fear the rise of Artificial Intelligence will
do; its goal is to teach students how to properly interact with human patients
when the time presents itself. Virtual Patients reduce the costs and do not require
the extensive number of resources necessary to hire and train human actors that
portray patients. These actors often misrepresent real life scenarios and data is
difficult to acquire because of the many variables natural to human interactions.
The Virtual Patients system bypasses these impedances by acquiring all the
necessary data while, on the surface, providing reasonable responses to their
proposed treatment.
Alfaro 7
The second of the projects at the University of Florida that border on recent
advancements in Artificial Intelligence is a smart house with the goal of aiding
the elderly that live alone. It is often the case that elderly individuals live alone
and when they experience a fall or a heart attack or some other sort of illness
there is no one around to come to their assistance. There are some technologies
that attempt to remedy the situation, such as alarms that individuals wear on
their wrists, but these tend to be very ineffective, for they require input from the
individual. The Gator Tech Smart House constantly monitors its resident(s) and
in the future will alert medical services if the need ever arises (Helal et al. 66).
This constant monitoring relieves the resident(s) from having to push an alarm or
call someone for help. Technologies such as the Smart House pose no threat. A
society in which Smart Houses house the elderly will eliminate the nursing homes
that so many of them loathe; they may hold on to their independence with the
assistance of technologies that allow them to help themselves.
The third and perhaps most useful of the research projects at the University of
Florida deals with an autonomous underwater submarine: SubjuGator. This
submarine was created for a competition of autonomous underwater vehicles and
won in 2005. Reliance in autonomous vehicles is necessary due to the value of
human lives; this need is made even more immediate by the great amount of
funding that the United States military has provided for such projects (Hoover
30). Autonomous agents, such as SubjuGator, can spare human soldiers in
military operations that require dangerous underwater missions for long periods
of time or explosives. These vehicles are being made increasingly smaller and
more powerful, soon able to be nearly undetectable. While SubjuGator is still in
Alfaro 8
its primitive stages and takes orders from its programmers, it exhibits the
beginnings an intelligent agent. Future autonomous vehicles will not rely solely
on algorithms, but will also identify how their environment and situations at
hand affect their actions. SubjuGator currently has the tasks of autonomously
navigating above pipelines, using its sonar to reach a specified location and
emerging from the water when it has repaired a broken pipe. An agent more
advanced than SubjuGator may, in the future, roam the waters close to the coasts
looking for drowning children and bring them to shore or protect aquatic life and
cleanse seawater. Of course, this agent would have to be intelligent and produce
results without human instruction. That is where Artficial Intelligence comes in.
Research in the field of Artificial Intelligence is common around the globe,
particularly the United States. At the Massachusetts Institute of Technology
(MIT) Kismet borders on what many hope is the beginning of artificial agents
that attempt to imitate human characteristics. Kismet is filed under what MIT
considers their “sociable” projects—the project’s goal is to facilitate human-robot
interaction. While Kismet does exhibit awkward pauses when trying to speak as
described above and fails to speak English, it is one of the most personable robots
today. Many humans forget that present day agents are still learning; humans are
afraid to start a conversation with Kismet for this reason. They may not know
what is happening or what the robot is “thinking” or attempting to say. But
Kismet in particular is equipped with sensors and motors that make its
movements very humanlike and make it approachable; these mechanisms allow
for the movement of eyebrows, forward and backward movement of its head,
movements of its eyes and ears (“Kismet”). Kismet may serve as a companion for
Alfaro 9
the elderly (along with Smart House), a child’s playmate, or a speaking agent at
an Automated Teller Machine (ATM).
Conclusion
Artificial agents have created unease in many; individuals fear that the human
race is straying from what is personal. Over reliance in computers is making
human interactions less frequent and the rise of Artificial Intelligence does not
help. Self-replicating and intelligent machines bring about fear that their
intelligence will reach levels far greater than that of the human mind and that
these machines will realize that humans are dispensable. From an evolutionary
standpoint, these machines are far superior than humans.
Recent research suggests that these phobias of new technologies are based on
ignorance. Humans are quick to jump to the negative aspects of these emerging
technologies and fail to see their useful applications in our society. Projects at the
University of Florida, Massachusetts Institute of Technology and other research
centers are proving that useful agents with positive goals are not only possible,
but being realized in the present day. The importance of these agents is
increasing. They are necessary to save human lives and increase the present
quality of living. The fear that intelligent agents may one day overthrow humanity
may be subdued only by the intelligence of those that create the artificial beings.
Engineers will have to use their mental abilities to equip these beings with
manual resets; humans should always be in charge.
Alfaro 10
With the use of artificial agents, surgeries may be performed more accurately
and more human lives may be saved. The benefits that arise from a society that
houses artificially intelligent agents are numerous and varied. The fear of these
agents will only hinder advancements that could improve life as it is currently
known.
Bibliography
Bicentennial Man. Dir. Chris Columbus. Perf. Robin Williams, Embeth Davidtz,
Sam Neill and Oliver Platt. Walt Disney Video, 1999.
Breazeal, C. Kismet. Massachusetts Institute of Technology’s Computer Science
and Artificial Intelligence Laboratory. 26 Feb. 2006
<http://www.ai.mit.edu/projects/humanoid-robotics-group/kismet/>.
---. “Robot in Society: Friend or Appliance?” Agents99. University of Washington,
Seattle. 1-5 May, 1999.
Drexler, K. Eric. Engines of Creation: The coming era of nanotechnology. New
York: Doubleday, 1986.
Helal, S. et al. “Gator Tech Smart House: A Programmable Pervasive Space.”
IEEE Computer Magazine. Mar. 2005: 64-74.
Hoover, Aaron. “UF’s winning robotic submarine provides peek at the future.”
University of Florida News. 29 Aug. 2005: 30.
Lok, Benjamin. VERG – Virtual Patients. 2 Feb. 2006. University of Florida. 24
Feb. 2006 <http://www.cise.ufl.edu/research/vegroup/vp.html>.
Perkowitz, S. Digital People: From Bionic Humans to Androids. Washington:
National Academies Press, 2004.
“Turing test.” Wikipedia: The Free Encyclopedia. 15 Feb. 2006. 25 Feb. 2006
<http://en.wikipedia.org/wiki/Turing_Test>.
Wolfe, Alan. “Mind, self, society, and computer: artificial intelligence and the
sociology of mind.” The American Journal of Sociology. 96(1991): 1073-96.