Artificial Intelligence
The Study of How Computers Could One Day Take Over the World
Jason A. Stubler
IT 103
George Mason University
Professor Karkehabadi
February 28, 2011
INTRODUCTION
Since the introduction of computer and information technology to our modern society, man has wondered to what extent these technologies could think for themselves. As time has progressed, several breakthroughs and advancements in the knowledge and practice of artificial intelligence have lead to the idea that computers may one day possess the ability to process thought at the same rate that humans are. With these advancements, however, come the inevitable concerns and debates that accompany any attempt to create an entity of super-intelligence. In this paper, I will analyze and discuss Watson, a new technological advancement developed by IBM, which was designed to answer questions correctly at a very high rate of speed.
BACKGROUND
A company known as International Business Machines, better known as IBM, created the artificially intelligent Watson. According to the official Q and A section of the IBM website, Watson is described as an artificially intelligent (AI), question answering computer system, named for IBM’s first president, Thomas J. Watson (2011). Additionally, according to an article in the Huffington Post, Watson’s internal system is powered by ten separate racks of IBM POWER 750 servers running a Linux operating system. It contains “fifteen terabytes of Random Access Memory (RAM), 2,880 processor cores and is capable of operating at 80 teraflops” (2011). According to computerworld.com research, the statement that Watson can operate at 80 teraflops means that Watson “only” operates at 80% of the processing power of the human brain (2011).
TESTING/JEOPARDY APPEARANCE
On Monday, February 14th, 2011, IBM tested their new creation, which specializes in answering questions in human language, by having it compete on the well-known ABC game show Jeopardy. The experiment had one match on the 14th, a day off, followed by a final match on Wednesday the 16th. Watson competed against two Jeopardy contestants, Ken Jennings and Brad Rutter, who are considered the best players to have played the game. Jennings is most famous for his winning seventy-four consecutive Jeopardy appearances, while Rutter holds the distinction of having won the most money ever on Jeopardy. According to Moore, the final tally, Rutter came in third overall with a three-day (two-match) total of $21,600, while Jennings finished marginally ahead of him with $24,000. Watson was victorious, obliterating its human opponents by having $77,147 to its name at the end of the competition (2011). According to a February 24th article in the New York Times by John Markhoff, the researchers who developed Watson did acknowledge the fact that Watson had one distinct flaw, saying the machine “benefited from the ‘buzzer factor’ ” (2011). This essentially means that Watson was programmed to ring in as quickly as possible, which gave it the advantage of being able to ring in first on the majority of questions.
POTENTIAL BENEFITS
With this advancement in artificial intelligence, a number of potential benefits have come to the forefront of possibility. One huge benefit of the development of artificial intelligence is that it allows for a much easier copying ability. When one machine is perfectly programmed, it becomes a simple and straightforward process to copy the information over and retain effectiveness in an equivalent machine. Human beings, on the other hand, are prone to the “learning curve” and are much more likely to make mistakes in the learning process. Another potential benefit would be the lack of need for sleep. If a machine that was artificially intelligent were to be put into a working environment where a specific task needed to be done, it would create a much more efficient production level. The machine would not require any level of rest, which would theoretically allow it to be able to work around the clock without interruption. Hoppe, Verdejo and Kay (2003) describe the benefits as “non-perishable and consistent. Where a human would eventually meet their end after a long period of life, machines can constantly work and be reproduced and reprogrammed quickly. The consistency of the machine’s work would go far beyond what an average human would be capable of” (p. 13). In the case of Watson, people could ask it a question at any given time during the day when an idea pops into their head, and it could be answered correctly by a machine that does not require sleep to think accurately and calculatingly.
SOCIAL PROBLEMS/ETHICAL ISSUES
Of course with any advancement in artificial intelligence comes the inevitable fears and ethical and social issues. A primary issue many critics seem to have with the development of artificial intelligence is the feeling that people will eventually be overtaken by machines. With the possibility of machines producing up to three times as quickly and as much as the average human worker, critics are worried that humans will no longer have a place in the working world, which could create a need for humans to replace their working hours with other activities. Another major issue with these creations is just what type of bias would be involved with its programming. According to Bostrom (1998), “The risks in developing super-intelligence include the risk of failure to give it the super goal of philanthropy. One way in which this could happen is that the creators of the super-intelligence decide to build it so that it serves only this select group of humans, rather than humanity in general” (pp. 12-17). Bostrom is essentially saying that a programmer could use their ability to program these machines to serve specific whims, rather than humanity as a whole. This becomes an issue because these artificially intelligent beings could be programmed to serve the betterment of a few, which would lead to drastically reduced conditions for those who may need the machine’s help a great deal more. Watson’s creation, which is purely to answer questions, could be altered in this way to have answers be biased and therefore incorrect. If one is to believe these incorrect answers, it could be, to a small extent, construed as a brainwashing technique.
FURTHER REQUIRED RESEARCH
As is the case with several forms of technology, research is still far from being completed. Scientists have only begun to scratch the surface of artificial intelligence, and have several avenues in which they can still explore. As previously mentioned, Watson can only currently operate at 80 teraflops, which is 80% of the average human brain’s 100 teraflop processing ability. With a general consensus among the human population that computers have the potential to outgrow humans in terms of intelligence, scientists have become further convinced that it is possible for computers to not only out-think a human, but to out-perform them in several additional aspects.
CONCLUSION
With future technologies constantly arising in the field of artificial intelligence, research is constantly ongoing and always changing. The benefits and risks appear to mostly equal out, so the true nature of what the improvement of artificial intelligence truly means for the human race is still a highly debated topic. It will continue to be this way for a number of years to come, and will remain that way until a definitive answer is discovered. Perhaps the next step would be to improve schools so that future children will be able to defeat the machines… in Jeopardy.
REFERENCES
Bostrom, N. (1998). Ethical Issues in Advanced Artificial Intelligence.
Retrieved from http://www.nickbostrom.com/ethics/ai.html
Computerworld (2011). Brain Behind IBM’s Watson Not Unlike a Human’s.
Retrieved from http://www.computerworld.com/s/article/print/9210319/Brain_behind_IBM_s_Watson_not_unlike_a_human_s?taxonomyName=Mainframes+and+Supercomputers&taxonomyId=67
Hoppe, U. & Verdejo, M. F., & Kay, J. (2003). Artificial Intelligence in Education: Shaping the
Future of Learning Through Intelligent Technologies. Amsterdam: IOS Press.
IBM (2011). Frequently Asked Questions.
Retrieved from http://www.research.ibm.com/deepqa/faq.shtml#3
Markhoff, J. (2011). Computer Wins on Jeopardy: Trivial, It’s Not. New York Times.
pp. 6B.
Moore, F. (2011). ‘Jeopardy’ Computer Crushes Competition. The Huffington Post.
pp. 1A, 2A.