Abstract
In the recent times, scientists in the discipline of artificial intelligence are confused which approach closely imitates the structure of a human brain. The best two approaches are bottom-up, or the top-down, and they comprise of both their advantages and disadvantages. The top-down approach has the capability to perform high-level tasks since it is programmed (Russell et.al 27). On the contrary, the bottom-up approach is capable of modeling the lower-level human functions such as motor control, image recognition, and learning capabilities. Each method operates perfectly where the other method is imperfect. Therefore, to establish the advantages and disadvantages of artificial intelligence, there is the need to perform more research. The Top-Down AI and the Bottom-Up AI are important approaches to artificial intelligence. The benefits accrued to AI has helped to improve the life of human beings because it can be applied in various sectors like education, military, astronomy, etc.
The definition of artificial intelligence
Artificial intelligence is a computer science discipline that gives machines the ability to appear like they have human cleverness. It can also be defined as the ability of an automated machine to imitate and simulate aspects of human behavior.
The history of artificial intelligence
The idea of inducing inanimate objects with intelligence have always been in existences over a long period. During the ancient days, the Greeks had mythologies about robots. Also, Egyptian and Chinese engineers has started to build automatons and machines (Lewis). To understand how AI started, it is good to understand the efforts made by classical philosophers’ to assess the thinking of human beings as symbolic systems. It is important to remember that the term, “Artificial Intelligence” was created at a conference at Dartmouth College, in Hanover, New Hampshire in 1956 (Lewis). This is the year that AI was formally founded
The period from 1974-80, also referred to as “AI winter” experienced inadequate funds from the government. Criticism about AI had grown to a large extent since developing an artificially intelligent being had proven tough. The field of Artificial Intelligence was refreshed in the 1980s after more funds from the British government as they aimed to compete with the Japanese (Lewis). Another major winter (1987-93) affected AI and coincided with the failure of the market for several general-purpose computers invented in the past, and this led to reduced funding from the government once again.
In the 1990’s, AI experienced major advancements in all areas including machine learning, multi-agent planning, scheduling, case-based reasoning, data mining, uncertain reasoning, vision, virtual reality, natural language understanding and translation, amongst others. The development of the humanoid robot in Rod Brooks Project at MIT also experienced significant progress (Lewis). Furthermore, a championship-level game-playing program was created with the help of a backgammon program developed by Gerry Tesauro, and it was named as TD-Gammon. It had the ability to compete favorably with world-class players.
1997 was a great year in the AI field, the Deep Blue Chess program defeated the world chess champion, NASA succeeded to land on the planet Mars and web crawlers and other essential programs spread throughout the universe. During the 2000’s, the field of AI experienced immense inventions. The interactive robot pets became sustainably commercially, and robots started expressing emotions with a face, and the Nomad Robot traveled in secluded Antarctica regions to search for meteorite samples.
How to categorize artificial intelligence (Top-Down AI and Bottom-Up AI)
Bottom-Up AI
The bottom-up category is very reasonable. The approach takes up the available assets stock and builds them up to increase the complex behaviors (Champandard). Bottom-up is beneficial because it applied the minimalism approach. It ensures that anything created is of use. It leads to the so-called “minimum length description” of the behaviors. The AI is responsible for managing the assets (Le Meur, Olivier, et al. 810). The problem with Bottom-Up AI is that it lacks the ability to create a mental model. Therefore, it becomes difficult to introduce different assets because the AI logical structure is weakened.
Top-Down AI
The approach is much more reliable. It commences by designing the game after which it is refined to match character AI level. The approach has its advantages in that it pays attention to AI parts that are vital to the design (Champandard). The AI comprises of a particular mental model that equals the design. The problem with this approach is that it has a tendency of ignoring the available resources at the start, and behaviors easy to create.
Thesis
Even though most scientists will be willing to see a generation of artificial intelligence and they will list a series of conveniences, there will be, anthropologists suggests that will also result in plenty of moral issues. To find the pros and cons of artificial intelligence, much more research will have to be done. The solution to this controversy is highly applied Top-Down AI, not Bottom-Up AI.
II. Positive aspects of Artificial Intelligence
Can they assist us to do some work we ask them to do
The Top-Down AI is significant because it helps machines with the intelligence to imitate and sense the ability of human beings to perform the assigned responsibilities. For instance, AI can help us to do some work we ask them to do. Although they lack real intelligence, they follow the instruction to perform the tasks, for instance, an intelligent vacuum cleaner. An intelligent vacuum cleaner adheres to a sequence of preprogrammed instructions to perform a task (Ulrich 240). It is assumed that the vacuum cleaner to be comprised of an inbuilt intelligence commands on what is supposed to be done and not what should be done. The intelligent vacuum cleaner is programmed to guide itself in a house picking up the dirt by use of spinning brushes and a vacuum. In the future, more programs will be created to assists people in various duties.
Can they process and compute by themselves?
Machines combined with artificial intelligence requires require computations if the scientists set the program. Additionally, the program is made up of a set of instructions that send commands and in the process, the inanimate objects can compute. For instance, the AlphaGo operates under this philosophy. AlphaGo is a computer game that operates under flexible algorithms brought about during its design. The application of artificial intelligence to the AlphaGo game has been used as a testing ground to solve problems similar to people. AlphaGo can extensively use the Google Cloud Platform installed by scientists to calculate a range of operations. Through the program, it acquired the capability of discovering new strategies by itself as it was involved in thousands of games. The game operates within its neural networks and helps to adjust its connections via reinforcement learning (trial-and-error process).
AlphaGo was tested and proven by scientists that it had more intelligence to defeat professionals. In addition, it is not a good idea to consider it an expert system comprised of hand-crafted rules, but individuals must always consider using the general machine learning techniques to compute processed data and engage in competition and go ahead to win it.
Are they always accurate?
When testing a sample, AI analyzes and compares it with the information stored in the database. As a result, it has all reasons always to be accurate. For instance, fingerprint detection. The fingerprints are taken when an individual presses the fingers on an ink pad after which the fingers are rolled on a paper leaving a clean image on the page. The police and security officers have always depended on the technology to identify the criminal individuals. Fingerprints scanners also operate with the help of AI and have proven to be effective. For instance, the scanners can be used to monitor the access to structures with the help of a computer system (Dash, Manoranjan, and Liu 155). Although it uses more sophisticated methods, the codes are placed in the database and matched to identify an individual. An optical scanner is helpful because it shines a bright light on the fingerprints and captures a digital photograph. The computer system will analyze the image and select the fingerprint in which by the help of an intellectual pattern-matching software that transforms it into a code.
Can they be highly applied?
Absolutely. For a long period of time, AIs have been applied by human beings in numerous fields. Their application benefits human beings in numerous ways. The AIs can reason and help in resolving issues that have been posing challenges to the existence of human beings. Even though AIs have the capacity to reason and solve our problems, people be reminded that they do not possess any level of self-awareness.
AIs can be applied in many aspects of life. For example, the use of the robotic systems have been applied in various disciplines to improve the condition of life. For instance, robots can be used as firefighters to reduce the risks that are exposed to civilians. Moreover, devices have been programmed to assist in performing certain tasks. This can be seen in the use of vacuum cleaner, lawn movers, dishwashers, and many others. AI has also been applied to improve the security in the society. For example, the use of alarm systems is important in fighting crime in the society. AI can be highly applied in various disciplines to improve the way of life.
Another application is that it is significant in boosting security at the national and individual levels. AI is used by various national security agencies and institutions. Nevertheless, the AI is associated with a high accuracy of output on solutions. The AI makes it easy to track the criminal profiles and research on their criminal acts.
Artificial intelligence is applied in the education sector and learning sector. For instance, it is used to offer personalized tutoring and assesses the student’s study patterns to forecast when they will become unaware and provide help at the level by notifying the tutor. Education software can be applied to suit the needs and wants of different students, provide feedback and automate arduous activities. The continuous automations is significant because it leaves the teacher to pursue teaching. The robots programmed with artificial intelligence facilitate the development of voice recognition systems. Nonetheless, AI is influential in the research and development in the medical disciplines.
III. Negative aspect of using Artificial Intelligence
Artificial Intelligence
The computers do not have a clue on anything and this implies that they not know that they do exist. Unlike the machines, the humans are self-aware that they are in existence and are aware of our surroundings and what is occurring around us. Self-awareness does arise from an information process in any way and will never be computer-generated in any mode. Awareness can only be quantified by awareness itself. The body and brain are apparent to be in awareness and that the self-awareness does arise from the body or brain. Nonetheless,, awareness does not refer to a physical thing and cannot be restrained by logical extremes.
Hard to imitate the structure of brain
Digital computers have the prowess to emulate the norms of other digital computers because computers work in a clearly stipulated feministic mode. For a computer to be simulated, the person is bound to perform the series of instructions that the computer being modelled would execute.
The human brain comprises of billions of neurons that functions with all the preceding data from our senses, emotions and innovations. The human nervous system comprises of simple neutrons which reveal that the human brain is a one large, sophisticated computer. It is complicated on certain aspects such as intelligence and cognition. The chemical elements as a result of generating an electric signal are some of the most complicated aspects about a neutron. The human brain has the capability to imitate a single neutron on a machine. Though, it difficult for the artificial intelligence to emulate the 1000 billion of neutrons that create intelligence and creativity (Lee).
Really hard to define intelligence
It is difficult to define intelligence. A computer could deceive a human into believing that it is a human it can be defined as intelligent. The discipline of Artificial Intelligence could give several definitions to the audiences. The terminology ‘Artificial Intelligence’ is so broad that people have categorized it into strong AI and weak AI. Additionally, the strong AI makes assumption that computers can be made to think on a level equal to humans. Moreover, artificial intelligence is the imitation of human intelligence processes by machines.
Ethical Concern
They will replace human. People will lose jobs.
Artificial intelligence is perceived to replace human jobs in the future. The world economic forum views that jobs will be lost via redundancy and automation. Computers could replace human jobs in customer service and call Centre roles. For instance, BMW has designed a system to allow people to answer questions within three seconds. The software that is known as BMW I Genius can comprehend a huge variety of questions and can reply with an accurate answer. The answer given by the software is indistinguishable from that given by a human expert.
How to treat an AI with self-awareness? Treat them like a real human being?
Artificial intelligence may sprout that some computer minds may achieve self-awareness. Artificial mind could develop interests on researching about certain topics. Artificial intelligence will address itself in a certain way that would develop in relation to its individual preferences. The AI tends to have a personality (McGovern). AI correlates to the interests and hobbies. In the period that the artificial intelligence fails to haunt for its interests and hobbies, the AI mind might be susceptible to remorsefulness. Machines need to be conferred anything that is presently reserved for human beings. However, it may be complicated to anticipate how the establishment of freedom may progress. The beginning of responsive machines will launch a new period for ethics, philosophers, lawyers and judges.
Will they attack human?
Nick Bostrom, Ray Kurzweil and Robin Hanson are view that the artificial intelligence does not pose any threat to human beings. They assert that the progress of the intelligence machines is a slow and gradual process. The computers that possess superior human intelligence and human beings depend on each other. They argue that superior intelligence necessitates a lot of experience. The theories perceive that the effectiveness of computational power to offer solutions to intellectual issues is extremely low. The machine have insufficient context to allow human beings to link naturally to one another. Another argument is that the machines heavily rely on human beings. Nearly all the machines that are in existence depend in humans to boost power, repair them after a malfunction and manufacture more when they their shelf life reduces (Lee).
Another reason is that relationships tend to get better under the power of AI. Bostrom believes that machines could become more powerful to the extent of controlling their preferences. However, putting in account the work of human societies, the intelligence has no sufficient room to become powerful. If this was possible, the human societies would be run by their philosophers or scientists. Therefore, the artificial intelligence will not attack the human beings.
In the contemporary world of intelligence, the most precious resources will be those that are normally restrained. These valuable resources are controlled by the human beings. The people will have at least control over intelligent computers as they will have on us beings.
IV. The solution
Bottom-Up AI
More research should be performed to examine the flexibility and adaptability of Bottom-Up AI in the human community before applying them. The bottom up approach applies parallel processing and advanced data structures. The approach encompasses artificial neural networks in simulation of the brain’s structure.
The bottom-up approach conduct analysis of thoughts that can be recreated in these artificial networks. The approach imitates the networks of artificial neurons that are equal to the neurons in the human brain.
Top Up AI
The top Up AI offers the solution
The top-up approach shows indications of true intelligence as the entire aspects of comprehension are placed on the logic of digital machines to understand. Humanity is likely to slow down if the prospects of progressing to be a super intelligent machine able to store information a person wanted to know. Strong AI is a combination of the top up and bottom-up approaches to produce effects (Hay). The top-up approach has demonstrated Impressive results that take the form of semantic comprehension. One common application is duplicating the logic based on the language to craft speech and language interpretation systems. In top-up AI, thought is treated as a high-level idea that is independent of the minimal details of the implementing mechanism.
V. Conclusion
There is the need to perform extensive research to understand the advantages and disadivantages of artificial intelligence. The solution linked to this controversy is associated with Top-Down AI and not Bottom-Up AI (Judd, Tilke, et al.). The top-down approach looks like it depicts the way in which human beings apply their knowledge in conversation. In situations where the top-down approach fails to solve issues, the bottom-up approach is applied. It is always important to note that AIs are not like human brains. They have the ability to learn but lacks the capability to manipulate and alter data in the manner at which a brain or any other programs subscribes to a top-down approach. The top-down approach has been cumbersome to be used by robotics since it needs a large amount of storage. As an alternative, the bottom-up approach is preferred. It has proven to be effective in different issues like face recognition, motor control, or any other human attributes. It is unfortunate that the bottom-up approach cannot handle situations with a higher-level complexity.
Work cited
Champandard, Alex J. "Top-Down Vs. Bottom-Up Design for Character AI | AiGameDev.com." AiGameDev.com | Your Online Hub for Game/AI. N.p., 24 July 2007. Web. 14 Apr. 2016.
Dash, Manoranjan, and Huan Liu. "Consistency-based search in feature selection." Artificial intelligence 151.1 (2003): 155-176.
GoogleInc. "Official Google Blog: AlphaGo: Using Machine Learning to Master the Ancient Game of Go." Official Google Blog. N.p., 26 Jan. 2016. Web. 14 Apr. 2016.
Hay, Maciamo. "Could a Machine or an AI Ever Feel Human-like Emotions ? H+ Media." H+ Media. H+ Media, 29 Apr. 2014. Web. 14 Apr. 2016.
Judd, Tilke, et al. "Learning to predict where humans look." Computer Vision, 2009 IEEE 12th international conference on. IEEE, 2009.
Le Meur, Olivier, et al. "A coherent computational approach to model bottom-up visual attention." IEEE Transactions on Pattern Analysis & Machine Intelligence 5 (2006): 802-817
Lee, Timothy B. "Will Artificial Intelligence Destroy Humanity? Here Are 5 Reasons Not to Worry." Vox. Vox, 29 July 2015. Web. 14 Apr. 2016.
Lewis, Tanya. "A Brief History of Artificial Intelligence." LiveScience.com. N.p., 4 Dec. 2014. Web. 14 Apr. 2016.
McGovern, Luke. "Artificial Intelligence - Top-down Vs Bottom-up Approaches to Strong AI." Flickspin - Online Writing Community. Flickspin, 15 Aug. 2008. Web. 14 Apr. 2016.
Russell, Stuart, Peter Norvig, and Artificial Intelligence. "A modern approach." Artificial Intelligence. Prentice-Hall, Egnlewood Cliffs 25 (1995): 27