Documentos de Académico
Documentos de Profesional
Documentos de Cultura
machines in the world that can think, that can learn and that can create. Moreover, their ability to do these things is going to increase rapidly until in a visible future- the range of problems they can handle will be coextensive with the range to which the human mind has been applied Herbert Simon
Ancient Times Aristotle: Logic is an instrument for studying (intelligence) Artificial Intelligence in Manufacturing
As the manufacturing industry becomes increasingly competitive, sophisticated technology has emerged to improve productivity. Artificial Intelligence in manufacturing can be applied to a variety of systems. It can recognize patterns, plus perform time consuming and mentally challenging tasks. Artificial Intelligence can optimize your production schedule and production runs.
Advantages
View your best product runs and the corresponding settings. Increase efficiency and quality by using optimal settings from past production. Artificial Intelligence can optimize your schedule beyond normal human capabilities. Increase productivity by eliminating downtime due to unpredictable changes in the schedule. The Tuppas Difference at Tuppas our focus is on continuous innovation. We provide your team with the ability to out innovate your competitors by providing tools that can affordably respond to your changing expectations on demand.
AI in Production Scheduling
Our artificial intelligence software for scheduling is based on genetic scheduling algorhythms which translate your scheduling goals into ordered tasks based on their importance. Tuppas Artificial Intelligence for Scheduling is designed to optimize your schedule based on your requirements. We design the software to recognize various levels of priority based on numerical associations. For example, if you are more concerned with a product or project due date than machine efficiency, but you want 1
the system to optimize both, we would program your system to give higher priority to the due date but still optimize for machine efficiency. Another example of AI software's capability would be when unplanned jobs need to be added to the schedule or an existing job changes priority; the system would immediately reorganize the entire schedule to include the new information while meeting your requirements and priorities.
Reduced IT software support requirement Reduced hardware and servers Intuitive, configurable system interfaces Reduction in software training Customizable open source code Cost of future innovation is dramatically decreased Ability to respond to new opportunities increases
During the past decade there has been a tremendous increase in the use of artificial insemination (AI) by commercial swine producers in the United States. In 1990, less than 7% of sows and gilts were bred by AI (Safranski, 1997). In contrast, a recent study of the structure of the Pork Industry revealed that nearly 50% of the nations sow herd was bred via AI in 1997 (PORK'98 Staff,
1998). The use of AI will continue to increase and within the next decade, nearly all swine producers will employ this technology. Producers utilizing AI can obtain semen in several ways: 1) Semen can be collected from boars housed on the farm or at farm-owned, off-site studs. 2) Semen can be obtained from studs that are structured so that the individual producer retains ownership of his boars. For a fee, the stud provides "room and board" for the boars as well as semen collection and processing services. 3) Some producers have joined together to create cooperative boar studs from which they obtain semen. 4) Semen can be purchased from commercial studs, the majority of which are located in the Midwest. With regard to obtaining semen from distant locations, overnight delivery companies such as UPS facilitate the use of AI for "mating" animals housed at widely separated locations. The method of obtaining semen a producer chooses depends on several factors including available labor, technical skills and facilities. Each method has advantages and disadvantages. For example, on-farm semen collection and processing allow a producer a readily available supply of fresh semen and complete control of genetic and semen quality decisions. However, specialized labor and equipment is necessary for semen collection and processing and capital is invested in boars. Obtaining semen from off-farm sources negates the need for the aforementioned equipment and labor. There is a potential, however, for semen delivery problems and the producer is at the mercy of someone else for many genetic and semen quality decisions. Dr. Don Levis at the University of Nebraska has developed computer spreadsheets to help determine which method is most cost-effective for a particular producer's situation. Most semen utilized in AI programs is in a liquid (fresh) form and is utilized within several days after collection. It has been welldemonstrated that farrowing rates and litter sizes achieved using liquid semen can be equal to or better than those resulting from natural mating systems (Flowers and Alhusen, 1992). Frozen semen, which can be stored almost indefinitely, is available from some commercial studs. Use of frozen semen, however, results in lower farrowing rates and smaller litter sizes when compared to those achieved using natural service or AI with fresh semen. AI offers producers numerous benefits. By allowing more extensive use of superior sires, AI can enhance the rate of genetic improvement and in terminal mating systems, increase the consistency of market hogs produced. The increased use of a genetically superior boar can be illustrated in an example. Semen could be collected from a boar 2 to 3 times a week without adversely affecting semen volume or sperm concentration. A 3
typical ejaculate could be diluted with extender in such a manner as to provide 12 insemination doses. Assuming the boar works 52 weeks a year and sows are bred twice while in estrus, the boar could "service" 624 females in a single year (2 collections per week x 12 doses of semen per collection x 52 weeks per year divided by 2 doses of semen per sow in estrus). The same boar used for natural mating could never service that number of females in one year. That AI decreases the number of boars needed by a farmer is an economic benefit because fewer boars mean less feed and maintenance costs. Moreover, higher prices for particularly good sires can be justified because fewer boars are needed. Again, an example is in order. With hand mating, a sow-to-boar ratio of 16:1 is fairly typical. Thus, for a 300-sow unit, 19 boars would be required. With an on-farm AI program, the sow-to-boar ratio is increased to 100:1 and only 3 boars are required. Assuming boar feed and maintenance costs of $1.00 per boar per day, the savings accrued by using AI in this example amounts to $16 per day or $5,840 dollars per year. Additionally, if the producer could spend $10,000 for boars, using AI he could buy 3 high-quality boars at $3,333 a piece. Using natural mating, the producer could be forced to purchase 19 mediocre boars at $500 a head. If a producer adopts an AI program for which semen is obtained from off-site sources, the only boars that must be housed on the farm are those used for estrus detection. Additionally, one or more boars are usually kept for "emergency" matings in case there is a problem with semen delivery. When obtaining semen (as opposed to boars) from outside sources there is a decreased risk of introducing diseases. Some diseases (e.g., Porcine Reproductive and Respiratory Syndrome [PRRS] can be transmitted via animal to animal contact as well as AI. Other diseases (e.g., pseudo rabies), however, cannot be transmitted through semen. Boars housed at studs are tested for, and vaccinated against, a variety of common swine diseases. Moreover, semen is collected under very sanitary conditions. Another advantage of AI is that it allows the mating of animals of different sizes. For example, older, larger boars can be "mated" to smaller gilts when employing AI. A final advantage of AI is that compared to hand-mating systems, it can be a time saver. Flowers and Alhusen (1992) indicated that any time more than four animals were bred in a day, the amount of labor required to supervise matings was significantly less with AI than for natural service. Additionally, as the number of matings per day increased, so did the relative labor advantage of AI over natural service. In that study, estimates of labor requirements for AI included collection and processing of semen, cleaning equipment, etc. Obviously, the amount of labor required for an AI 4
program in which semen is purchased, rather than collected on the farm, is even less. Two final comments should be made regarding AI. First, it certainly is not a requirement that breeding programs be exclusively natural or all AI. In fact, many producers effectively use a combination of natural mating and AI. For example, producers may use natural mating for the first service of a sow in estrus. Second and possibly third matings are then accomplished via AI. This system also works well as a "transition" system for producers gradually moving from all natural to all AI breeding programs. Secondly, AI is a reproductive technology that can work for all producers regardless of the size of their operations.
Achievements of AI
Deep Thought is an international grand master chess player. Sphinx can recognise continuous speech without training for each speaker. It operates in near real time using a vocabulary of 1000 words and has 94% word accuracy. Navlab is a truck that can drive along a road at 55mph in normal traffic. Carlton and United Breweries use an AI planning system to plan production of their beer. Robots are used regularly in manufacturing. Natural language interfaces to databases can be obtained on a PC. Machine Learning methods have been used to build expert systems. Expert systems are used regularly in finance, medicine, manufacturing, and agriculture
Must be able to perceive its environment. Must be able to affect its environment. Must be able to reason about observations and actions Must be able to learn from observations and actions. Must have goals.
alone, even more are needed before the technology can be marketed and put to real use. According to the Pentagon, actual robotic soldiers powered by Artificial Intelligence will be a major fighting force in the American army, probably within the next decade. The first robot soldiers will actually be remote-controlled vehicles. The military has poured tens of billions of dollars into this project already. Congress wants to see this happen and they ordered that a third of all military vehicles and deep-strike aircraft be automated by 2010. As the machines begin to think, see, and react more like humans, the level of their autonomy and our level of trust in them will grow as well. However, it is predicted that a true soldier-simulating robot will not come about for another 30 years. These robots need to be able to determine friend from foe and enemy from bystander, and teaching them to do so will require a tremendous amount of research and work. The government has assured us however that these robotic soldiers will not be put into the field and allowed to make such decisions until they are ready to do so. Another current infantry prototype knows how to recognize an enemy when it is under fire. When this happens, it can react to enemy fire on its own or follow orders given to it from a remote observer. Although it's programmed to work autonomously, in its present state, it still requires some set of outside monitoring controls in order for it to work. Its designers plan to have it usable for infantry missions by 2015. Another one of their prototypes nearly realizes the anthropomorphic goal imagined by Isaac Asimov in his I, Robot book. This prototype is a machine about four feet high with a Cyclops eye and a gun for a right arm. It is programmed to perform basic hunting and killing tasks. It can actually find valid targets on its own and can shoot at them with remarkable accuracy. The list of benefits of using machines to achieve military goals is long and significant. The immediate and most evident boon of such technology is the elimination of human risk: machines, not
humans, would be lost in battle. In addition, specialized robots can be designed to accomplish specific tasks more effectively than humans can, increasing the military's overall effectiveness. They are also more cost-effective. Robots will always be able to do what they were designed to do and can be recycled when they are obsolete. A human soldier costs on average $4 million dollars over his lifetime, and the U.S. Pentagon cannot obtain the money to pay all of them. Robots could cost a tenth of that amount or less. Although the ultimate goal of the robot soldier is to completely eliminate human risk, even the experts say that war will always be a human endeavour involving human loss of life, no matter how much the AI warrior is developed. New ethical questions will arise once we have the ability to invade countries without risk of bloodshed on the part of the invader. And even though these robotic developments will soon be on our doorstep, its a little frightening to see that the only ones who are addressing the issue of use and or misuse of such technology are the scientists and the authors of science-fiction.
spacecraft fairly easily. As software intricacy increases, programs like Livingstone will find use in other, more common fields as well. In addition, Artificial Intelligence can use satellite pictures and mathematical models to look at the ocean and better predict the weather. Visible masses of water, or fronts, have distinct properties that make each mass different from the others. The ability of neural networks to deal with imprecise data makes them best able to deal with the unpredictable movements and temperatures of the fronts, though they still don't work over a long period of time. The computer can collect raw data such as the sea's density, salinity, and temperature at any given spot, from sensors placed in the ocean. With these, it can put together a model of the ocean and its movements, predicting where each front will move next. The ultimate goal of this project is to make a computer able to predict continuously the weather at any point over the ocean and without human intervention. Satellite observations of the worlds oceans are created with lasers, radars, and infrared scanning devices. This can be somewhat difficult as the electromagnetic waves cannot penetrate the water well, but this is overcome with data recorded from the upper layers of the oceans and provides aid to oceanographers in the navigation of shipping vessels. The first satellite, which aided in providing oceanographers with a map of the ocean floor, was the Seasat-A. This was created in the 1970s, and it produced a flight path in which it was able to circumnavigate the earth over a period of time, developing a visible picture of the ocean floor. The seafloor height was measured every few kilometres, and the data complied by the satellite was used to compile a global map of the surface of the sea and the seafloor. The trenches and ridges on the surface of the seafloor have an effect on the gravitational effects of the sea surface. The future of physical oceanography was greatly increased with the establishment of Seasat-A.
Robot Vision
When a camera is attached to a robot, the robot simply sees a bunch of data. With the camera picture alone, it has difficulty recognizing individual objects or patterns. However, with a little bit of AI technology and about $2 worth of off-theshelf parts, we can turn an ordinary camera into a device that a robot could use to discern still or moving objects, as well as bodies on a darkened street or in a smoke filled room. This all takes place through a process known as segmentation where a computer is used to split apart an image into various segments. For instance, a camera could flash an invisible infrared LED light over an area in its field of vision. Then, based on the brightness of the pixels returned, the machine attached to the camera could determine what object in its viewing area it wants to look at. Using such a segmentation camera, it could not only pick out human beings but even pick out hands, heads, or even more detailed entities for that matter. Motion and colour could also be used to segment an image. Having found the object its looking for; the computer could either follow its movement or continue its search for similar objects. Seg cameras can even recognize gestures, facial expressions and hand motions. A computer program like this, along with a limited speech recognition program, can perform simple tasks like hands-free Web surfing (using motion to scroll back and forth and speech to select links and menus). It can also be used to highlight and clarify a picture. For example, in a web conference, one is only interested in the clarity of the other persons face. A seg cam can isolate the face and make sure its clear, ignoring the rest of the image. A more advanced seg cam can theoretically read lips, and with speech recognition added in to the mix, it could function as a fairly accurate speech synthesizer. It could also investigate facial
10
features, giving audible warnings to drivers who show signs of falling asleep (they blink too long, yawn too much, or begin to nod off).
11
selecting from a handful of reasoning methods. The practice of knowledge engineering is described later. We first describe the components of expert systems.
reasoning. If the chaining starts from a set of conditions and moves toward some conclusion, the method is called forward chaining. If the conclusion is known (for example, a goal to be achieved) but the path to that conclusion is not known, then reasoning backwards is called for, and the method is backward chaining. These problem-solving methods are built into program modules called inference engines or inference procedures that manipulate and use knowledge in the knowledge base to form a line of reasoning. The knowledge base an expert uses is what he learned at school, from colleagues, and from years of experience. Presumably the more experience he has, the larger his store of knowledge. Knowledge allows him to interpret the information in his databases to advantage in diagnosis, design, and analysis. Though an expert system consists primarily of a knowledge base and an inference engine, a couple of other features are worth mentioning: reasoning with uncertainty, and explanation of the line of reasoning. Knowledge is almost always incomplete and uncertain. To deal with uncertain knowledge, a rule may have associated with it a confidence factor or a weight. The set of methods for using uncertain knowledge in combination with uncertain data in the reasoning process is called reasoning with uncertainty. An important subclass of methods for reasoning with uncertainty is called "fuzzy logic," and the systems that use them are known as "fuzzy systems." Because an expert system uses uncertain or heuristic knowledge (as we humans do) its credibility is often in question (as is the case with humans). When an answer to a problem is questionable, we tend to want to know the rationale. If the rationale seems plausible, we tend to believe the answer. So it is with expert systems. Most expert systems have the ability to answer questions of the form: "Why is the answer X?" Explanations can be generated by tracing the line of reasoning used by the inference engine (Feigenbaum, McCorduck et al. 1988). The most important ingredient in any expert system is knowledge. The power of expert systems resides in the specific, high-quality knowledge they contain about task domains. AI researchers will continue to explore and add to the current repertoire of knowledge representation and reasoning methods. But in knowledge resides the power. Because of the importance of knowledge in expert systems and because the current knowledge acquisition method is slow and tedious, much of the future of expert systems depends on breaking the knowledge acquisition bottleneck and in codifying and representing a large knowledge infrastructure.
13
Knowledge engineering
Is the art of designing and building expert systems, and knowledge engineers are its practitioners? Gerald M. Weinberg said of programming in The Psychology of Programming: "'Programming,' -- like 'loving,' -- is a single word that encompasses infinitude of activities" (Weinberg 1971). Knowledge engineering is the same, perhaps more so. We stated earlier that knowledge engineering is an applied part of the science of artificial intelligence which, in turn, is a part of computer science. Theoretically, then, a knowledge engineer is a computer scientist who knows how to design and implement programs that incorporate artificial intelligence techniques. The nature of knowledge engineering is changing, however, and a new breed of knowledge engineers is emerging. We'll discuss the evolving nature of knowledge engineering later. Today there are two ways to build an expert system. They can be built from scratch, or built using a piece of development software known as a "tool" or a "shell." Before we discuss these tools, let's briefly discuss what knowledge engineers do. Though different styles and methods of knowledge engineering exist, the basic approach is the same: a knowledge engineer interviews and observes a human expert or a group of experts and learns what the experts know, and how they reason with their knowledge. The engineer then translates the knowledge into a computer-usable language, and designs an inference engine, a reasoning structure, that uses the knowledge appropriately. He also determines how to integrate the use of uncertain knowledge in the reasoning process, and what kinds of explanation would be useful to the end user. Next, the inference engine and facilities for representing knowledge and for explaining are programmed, and the domain knowledge is entered into the program piece by piece. It may be that the inference engine is not just right; the form of knowledge representation is awkward for the kind of knowledge needed for the task; and the expert might decide the pieces of knowledge are wrong. All these are discovered and modified as the expert system gradually gains competence. The discovery and cumulating of techniques of machine reasoning and knowledge representation is generally the work of artificial intelligence research. The discovery and cumulating of knowledge of a task domain is the province of domain experts. Domain knowledge consists of both formal, textbook knowledge, and experiential knowledge -- the expertise of the experts.
14
15
In the early 1970s another AI programming language was invented in France. It is called PROLOG (Programming in Logic). LISP has its roots in one area of mathematics (lambda calculus), PROLOG in another (first-order predicate calculus). PROLOG consists of English-like statements which are facts (assertions), rules (of inference), and questions. Here is an inference rule: "If object-x is part-of object-y then a component-of object-y is object-x." Programs written in PROLOG have behaviour similar to rule-based systems written in LISP. PROLOG, however, did not immediately become a language of choice for AI programmers. In the early 1980s it was given impetus with the announcement by the Japanese that they would use a logic programming language for the Fifth Generation Computing Systems (FGCS) Project. A variety of logicbased programming languages have since arisen, and the term prolog has become generic.
set of actions to achieve those goals, and/or provide a detailed temporal ordering of those actions, taking into account personnel, materiel, and other constraints. This class has great commercial potential, which has been recognized. Examples involve airline scheduling of flights, personnel, and gates; manufacturing jobshop scheduling; and manufacturing process planning.
Knowledge Publishing
This is a relatively new, but also potentially explosive area. The primary function of the expert system is to deliver knowledge that is relevant to the user's problem, in the context of the user's problem. The two most widely distributed expert systems in the world are in this category. The first is an advisor which counsels a user on appropriate grammatical usage in a text. The second is a tax advisor that accompanies a tax preparation program and advises the user on tax strategy, tactics, and individual tax policy.
A speed-up of human professional or semi-professional work -- typically by a factor of ten and sometimes by a factor of a hundred or more. Within companies, major internal cost savings. For small systems, savings are sometimes in the tens or hundreds of thousands of dollars; but for large systems, often in the tens of millions of dollars and as high as hundreds of millions of dollars. These cost savings are a result of quality improvement, a major motivation for employing expert system technology. Improved quality of decision making. In some cases, the quality or correctness of decisions evaluated after the fact show a ten-fold improvement. Preservation of scarce expertise. ESs is used to preserve scarce know-how in organizations, to capture the expertise of individuals who are retiring, and to preserve corporate know-how so that it can be widely distributed to other factories, offices or plants of the company. Introduction of new products. A good example of a new product is a pathology advisor sold to clinical pathologists in hospitals to assist in the diagnosis of diseased tissue.
18
The ability to perform complex tasks without making human-type mistakes, such as mistakes caused by lack of focus, energy, attention or memory. The ability to perform extended tasks at greater serial speeds than conscious human thought or neurons, which perform approx. 200 calculations per second. Computing chips (~2 GHz) presently have a 10 million to one speed advantage over our neurons. The in principle capacity to function 24 hours a day, seven days a week, 365 days a year. The human brain cannot be duplicated or re-booted, and has already achieved optimization through design by evolution, making it difficult to further improve. The human brain does not physically integrate well, externally or internally, with contemporary hardware and software. The non-existence of boredom when performing repetitive tasks. Advantages of human brains over hypothetical AIs: Present AIs lack human general intelligence and multiple years of real-world experience. The computational capacity of the human brain is estimated at 2 * 10^16, or 20 million billion calculations per second, which is twenty times greater than the supercomputer Blue Genes predicted achievement of 10^15, or 1 million billion calculations per second, by 2005. However, the human brain may not have a computational advantage over computers for much longer. Ray Kurzweil, for example, predicts that the computational capacity of the human brain will be accomplished on supercomputers, or clustered systems, by 2010, followed on personal computers by 2020. The human brain has already achieved a high-level of complexity and optimization through design by evolution, and thus has proven functionality. Advantages of minds-in-general (AIs) over the human brain:
19
(The following are not advantages of specific AI approaches, but rather advantages of minds-in-general over the human brain.) An increased ability to acquire, retrieve, store and use information on the Internet, which contains most human knowledge. Lack of human failings that result from complex functional adaptations, such as observer-biased beliefs or rationalization. Lack of neurobiological features that limit human control over functionality. Lack of complexity that we have acquired from evolutionary design, e.g., unnecessary autonomic processes and sexual reproduction. The ability to advance on the design of evolution, which is continually constrained by lack of foresight, the requirement to maintain pre-existing design, and a weakness with simultaneous dependencies. The ability to add more computational power to a particular feature or problem. This may result in moderate or substantial improvements to pre-existing intelligence. (AI does not have an upper limit on computational capacity; we do.) Note that the speed of computational power is predicted to continually increase exponentially, and decrease exponentially in cost, every 12-24 months, in accordance with Moores Law. The ability to analyze and modify every design level and feature. The ability to combine autonomic and deliberative processes. The ability to communicate and share information (abilities, concepts, memories, thoughts) at a greater rate and at a greater level of complexity than us. The ability to control what is and what is not learned or remembered.
20
The ability to create new modalities that we lack, such as a modality for code, which may improve the AIs programming ability-by making the AI inherently native to programming - far beyond our own (a modality for code may allow the AI to perceive its hardware machine code, i.e. the language used to write the AI, and other abilities). The ability to learn new information very rapidly. The ability to consciously create, analyzes, modify, and improve abilities, concepts, or memories. The ability to operate on computer hardware that has powerful advantages over human neurons, such as the ability to perform billions of sequential steps per second. The capacity to self-observe and understand on a fine-grained level that is impossible for us. AIs may have an improved capacity for introspection and manipulation, such as the ability to introspect and manipulate code, which would be the functional level comparable to human neurons, which we cant think about or manipulate. The most important and powerful capacity of minds-in-general over the human brain is the ability to recursively self-encapsulate and self-improve its intelligence. As a mind becomes smarter, the mind can use its intelligence to improve its design, thereby improving its intelligence, which may allow further improvements to its design, thus allowing further improvements to its intelligence. It is unknown when open-ended self-improvement may begin.
21
If you think about it, we take in numerous pieces of information just walking down the street, much of it unconsciously. You might be thinking about the weather, the pace of your steps, where to step next, the movement of other people, smells, sounds, the distance to the destination, the effect of the environment around you and so forth. An artificial intelligence in a virtual world has fewer of these variables to deal with because as of yet, no virtual world approaches the complexity of the real world. It may be that by simplifying the world in which the artificial intelligence operates (and by working in a self-contained world), some breakthroughs can be achieved. Such a process would allow for a more linear development of artificial intelligence rather than an attempt to immediately jump to lifelike robots capable of learning, reason and self-analysis. Goertzel states that a virtual world also offers the advantage of allowing a newly formed artificial intelligence to interact with thousands of people and characters, increasing learning opportunities. The virtual body is also easier to manage and control than that of a robot. If an AI-controlled parrot seems to have particular challenges in a game world, it's less difficult for programmers to create another virtual animal than if they were working with a robot. And while a virtual world AI lacks a physical body, it displays more complexity (and more realism) than a simple AI that merely carries on text-based conversations with a human. Novamente claims that its system is the first to allow artificial intelligences to progress through a process of self-analysis and learning. The company hopes that its AI will also distinguish itself from other attempts at AI by surprising its creators in its capabilities -- for example, by learning a skill or task that it wasn't programmed to perform. Novamente has already created what it terms an "artificial baby" in the Ageism virtual world. This artificial baby has learned to perform some basic functions. Despite all of this excitement, the AI discussed here are far from what's envisioned in "Terminator." It will be some time before AIs are seamlessly interacting with players, impressing us with their cleverness and autonomy and seeming all too human. Even Philip Rosedale, the founder of Linden Labs, the company behind "Second Life," has warned against becoming caught up in the hype of the supposedly groundbreaking potential of these virtual worlds. But "Second Life" and other virtual worlds may prove to be the most valuable testing grounds to date for AI. It will also be interesting to track how virtual artificial intelligences progress as the virtual worlds they occupy change and become more complex. Besides acting as an incubator for artificial intelligence, "Second Life" has already been an important case study in the development of cyber law and the economics and legality of hawking virtual goods for real dollars. The popular virtual world has even been
22
mentioned as a possible virtual training facility for children taking emergency preparedness classes.
LNM
+$190 +$69 +$121
Genetics are cumulative. If you have been using a herd bull for multiple generations the differences between the second generation animals of A.I. versus natural service offspring is greater than the first generation. This difference increases as more generations add up. Table 2, looks at the expected difference in milk production and LNM$ after using an average A.I. sire rather than a herd bull for one, two and three generations. These calculations assume both groups start with an average cow and that genetic progress per generation is +200 pounds milk and +$90 LNM.
23
Table 2. Additional performance of generational use of average A.I.-sired cattle versus herd bulls.
Generation
Generation One Generation Two Generation Three
Milk LNM
+838 +1257 +1467 +$121 +$181 +$212
With the use of 80th percentile or higher LNM sires the advantage of A.I. versus natural service bulls after three generations more than doubles the advantage seen in Table 2. The bottom line is the advantages of A.I. are well documented. Besides the benefits of enhanced conception ability, safer working conditions, reduced insurance and higher milk checks, A.I. leads to superior genetics and more overall profitability.
Artificial Insemination
The advantages are many starting with the fact that through artificial insemination the improvement of a particular trait (like our example of Standard bred trotting/pacing speed) in the breed of horse can be accomplished comparatively quickly. This is due in part to the ability of a stud to impregnate many more mares per season than is possible with live cover. With a wider results base each season, the desired traits are more likely to show and are more likely to be rebred, therefore producing a quicker time frame of breed improvement. What then adds to the process is that if these offspring and only these offspring showing the improvement are selected for stud (as in the case of Standard bred breeding) the chances of the improved trait proliferating quickly are even higher. Other physical and management advantages of AI include the reduced risk of the spread of both venereal diseases and nonvenereal disease such as equine influenza, strangles and equine infectious anemia.One interesting and helpful break-through is the ability to treat semen to reduce the natural bacteria content, therefore reducing the risk of uterine infection in prone mares. long with the ability to impregnate mares with physical abnormalities (although, if genetic it is generally thought best not to breed these mares), the risk of injury that can occur with violent mares or stallions is also completely removed. And of course the last major advantage of AI is that you don't need to live down the street from the stallion or at least within practical hauling distance. AI makes it possible to breed your mare to practically any stud in the same hemisphere without so much as trying to load your mare into a trailer, not to mention avoiding the mare care fees that often
24
stack up at stud farms, especially if your mare doesn't take immediately. The use of artificial insemination can be very advantageous to the stallion owner as well because he reduces the risk of injury to his stud and also increases his profits since one ejaculation can breed multiple mares.
25
workload on a pilot. Modern pilots work in incredibly complex electronic environments - receiving information not only from their own radar, but from many others (principle behind JSTARS). Not only is the information load high, the multi-role aircraft of the 21st century have highly complex avionics, navigation, communications and weapon systems. All this must be organized in a highly accessible way. Through voice-recognition, systems could be checked, modified and altered without the pilot looking down into the cockpit. Expert/advisory systems could predict what the pilot would want in a given scenario and decrease the complexity of a given task automatically. Aside from research in this area, various paradigms in AI have been successfully applied in the military field. For example, using an EA (evolutionary algorithm) to evolve algorithms to detect targets given radar/FLIR data, or neural networks differentiating between mines and rocks given sonar data in a submarine. I will look into these two examples in depth below.
Neural-networks
Neural networks (NN) are another excellent technique of mapping numbers to results. Unlike the EA, though, they will only output certain results. A NN is normally pre-trained with a set of input vectors and a 'teacher' to tell them what the output should be for the given input. A NN can then adapt to a series of patterns. Thus, when feed with information after being trained, the NN will output the result whose trained input most closely resembles the input being tested. This was the method that some scientists took to identify sonar sounds. Their goal was to train a network to differentiate between rocks and mines - a notoriously difficult task for human sonar operators to accomplish. The network architecture was quite simple, it had 60 inputs, one hidden layer with 1-24 inputs, and two output units. The output would be <0, 1> for a rock and <1, 0> for a mine. The large amount of input units was to incorporate 60 normalized energy levels of frequency bands in the sonar echo. What this means is that a sonar echo would be detected, and subsequently fed into a frequency analyzer, that would break down the echo into 60 frequency bands. The various energy levels of these bands was measured, and converted into a number between 0 and 1. A few simple training methods were used (gradient-descent), as the network was fed examples of mine echoes and rock echoes. After the network had made its classifications, it was then told whether it was correct or not. Soon, the network could
26
differentiate as good as or better than its equivalent human operator. The network had also beaten standard data classification techniques. Data classification programs could successfully detect mines 50% of the time by using parameters such as the frequency bandwidth, onset time, and rate of decay of the signals. Unfortunately, the remaining 50% of sonar echoes do not always follow the rather strict heuristics that the data classification used. The networks power came in its ability to focus on the more subtle traits of the signal, and use them to differentiate.
Genetic Programming
Genetic programming is an excellent way of evolving algorithms that will map data to a given result when no set formula is known. Mathematicians/programmers could normally find algorithms to deal with a problem with 5 or so variables, but when the problem increases to 10, 20, 50 variables the problem becomes close to impossible to solve. Briefly, how a GP-powered program works is that a series of randomly generated expression trees are generated that represent various formulas. These trees are then tested against the data, poor ones discarded, good ones kept and breed. Mutation, crossover, and all of the elements in genetic algorithms are used to breed the 'highest-fitness' tree for the given problem. At best, this will perfectly match the variables to the answer, other times it will generate an answer very close to the wanted answer. (For a more in-depth look at GP, read the case study) A notable example of such a program is SDI's e evolutionary algorithm designed by Steve Smith. e has been used by SDI to research algorithms to use in radars in modern helicopters such as the AH-64D Longbow Apache and RAH-66 Comanche. e is presented with a mass of numbers generated by radar and perhaps a low-resolution television camera, or FLIR (Forward-looking Infra-red) device. The program then attempts to find (through various evolutionary means) an algorithm to determine the type of vehicle, or to differentiate between an actual target and mere "noisy" data. Basically, the EA is fed with a list of 42 different variables collected from the two sensors, and then a truth value specifying whether the test data was clutter or a target. The EA then generates a series of expression trees (much more complicated than those normally used in GP programs). When new a best program is discovered, the EA uses a hill-climbing technique to get the best possible result out of the new tree. Then, the tree is subjected to a heuristic search to optimize the tree. Once the best possible tree is found, e will output the program as either pseudo code, C, FORTRAN or Basic. While the algorithms performed well on the training data, the performance increased a lot when applied to the test data. 27
Nevertheless, the fused detection algorithm (using both radar and FLIR information) still provided a decent error percentage. An additional plus to this technique is that the EA could be actually programmed into the weapon systems (not just the algorithm outputted), so that the system could dynamically adapt to the terrain, and other mission-specific parameters.
Author/Futurist/Inventor Kurzweil Tech, Inc. Wellesley Hills, MA "You know, we look at a face and there's all kinds of complex calculations involved in recognizing and we do it instantly, and ultimately, our machines will have equal and, in fact, even greater powers of pattern recognition." He predicts as we reach a greater understanding of the brain, AI will advance even more. Ray Kurzweil "We'll be able to essentially recreate the powers of human intelligence and combine them with the speed, accuracy and knowledge-sharing ability of machines." Emotion is one quality still separating man from machine. But Eric Chown is working to narrow even this gap. He's reprogramming Aibo to feel emotions. A touch on his face is pleasure; his ear pushed forward - pain. His behaviours change with each emotion. Eric Chown, Ph.D. Computer Scientist Bowden College Brunswick, ME "I'm trying to show that this emotional system can be understood and we can start to make some real progress in learning how to deal with people who are emotional." This could prove helpful in treating conditions like Alzheimer's disease. Eric Chown, Ph.D. "What we would like to do is understand what's going wrong in that patient's head." At MIT, Charlie Kemp is programming his computer to have common sense. Charlie Kemp "Common sense is not something we as people can teach because it's all automatic for us." He predicts we will eventually go beyond humans teaching computers. Charlie Kemp MIT Artificial Intelligence Lab Cambridge, MA 29
"We will be able to work better together, whereas now it's pretty much humans, you know, slamming away at some computer, trying to get it to do something." In the future these experts predict humans and machines will actually merge. Humans will think using non-biological intelligence. Ray Kurzweil "We'll have billions of nanobots in the capillaries of our brains, communicating wirelessly with our biological neurons, with the Internet, with each other, and basically expanding human intelligence and experience." Kurzweil calls it evolution. Ray Kurzweil "It's not some alien invasion of intelligent machines. It's coming from within our civilization. It expands our own intellectual powers." Just as the primates probably could not have predicted our civilization, few of us can imagine what evolution will bring next. Kurzweil predicts the merger of our intelligence with artificial intelligence will happen as soon as the 2020's, and perhaps by 2030, the artificial portion of our brains will dominate. He points out the merger has already begun as people with Parkinson's disease are being treated with electrodes implanted in the brain. BACKGROUND: Today, artificial intelligence helps airplanes fly, makes financial decisions, and helps diagnosis medical conditions. Tom Mitchell, president of the American Association for Artificial Intelligence, says this about AI: "Ever since computers were invented, it has been natural to wonder whether they might be able to learn. Imagine computers learning from medical records to discover emerging trends in the spread and treatment of new diseases, houses learning from experience to optimize energy costs based on the particular usage patterns of their occupants, or personal software assistants learning the evolving interests of their users to highlight especially relevant stories from the online morning newspaper." The AAAI describes artificial intelligence as "the scientific understanding of the mechanisms underlying thought and intelligent behaviour and their embodiment in machines." Experts say AI is going to be increasingly important in our lives and it won't be long before AI allows man to increase his levels of intelligence. AI IN THE FUTURE: Author, inventor and futurist Ray Kurzweil divides AI into two types, narrow AI and strong AI. He says what we have now is narrow AI. Strong AI, he says, is "machines that can emulate the full range of diversity and subtlety of human 30
intelligence I feel that within 25 years, we will really understand how the human brain works. We'll have very powerful computers that can recreate those processes." He says AI will be able to expand the reaches of human intelligence and we're going to literally merge with our machines. He anticipates, in the future, people will have nanobots (tiny robots) in their brains that will allow them to communicate with the Internet and with each other. Also, these will allow thinking to be done at a rate that is impossible now, given the limitations of our brains. The human brain works using interneuron connections that have a signalling speed of 200 calculations per second speed 100 million-times slower than electronic circuits. "Ultimately," says Kurzweil, "when computers can emulate human forms of intelligence, they'll be able to combine it with this tremendous speed." AI IN MEDICINE: Currently, AI is already being used to help treat some medical conditions. For example, many patients with Parkinson's disease can be considered cyborgs because they have neural implants that replace the damaged biological neurons and these communicate with healthy biological neurons. Currently, this must be done with surgery but Kurzweil expects intelligent machines will be introduced through the bloodstream in the future. Eric Chown, Ph.D., at Bowden College in Brunswick, Maine, is working to teach emotions to a robot one characteristic still separating man from machine. The goal is to get a better understanding of how emotions impact behaviours and eventually use this information to help people with dementia or Alzheimer's disease function better in their environment. Sure, his ear pushed forward - pain. His behaviours change with each emotion.
previously, each of the seven churches represents a cell assembly (sub network) in the brain. In turn, each cell assembly has a unique function and architecture and performs a specific signal processing task which is essential to the functioning of the system as a whole. Using this knowledge, it is possible to endow a machine with the ability to learn from its environment and attain sophisticated behavioural and reasoning skills, very much like a human being. Intelligent machines will run the gamut from self-piloting ground and air vehicles to highly advanced robotic systems. They will become proficient in every human field of knowledge and expertise. There will be robot doctors, nurses, engineers,
31
technicians, scientists, chefs, soldiers, historians, construction workers, farmers, architects, store managers, investment advisors, language interpreters, gardeners, housekeepers, landscapers, factory workers and many more. They will come in all shapes, sizes and specialties. Some will move about on wheels, some will fly, and others will sport multiple legs and arms and various other appendages. Many will interact directly with humans while others will remain hidden from view most of the time, coming out to do their chores quietly when no humans are in sight. Some will not move at all, as their function will be to search the internet and databases for new knowledge. But this is just the tip of the iceberg because, for the first time in recorded history, humanity is in a position to build machines so powerful, that their intelligence will surpass that of the smartest human beings on earth! And contrary to what some experts may suppose, it can be done now, using existing computer technology. Where will this road take us? It is anyone's guess. But this is undoubtedly the dawning of the age of the super intelligent machine.
32
change to a better system? Or will we plunge headlong into a global Orwellian Nightmare? Here is a hint: The solutions to all these problems can all be found in the same book where the secret of intelligence came from. It has been said that AI is the ultimate invention, i.e., the invention to end all inventions. It is easy to envision a golden age in which humanity's every need and desires are met by legions of intelligent and benevolent synthetic servants. And why not? The development of truly intelligent machines can certainly bring about an age of plenty and careless living for all. Furthermore, super-intelligent machines will be hard at work searching for solutions to some of the world's most pressing problems. However, the road to utopia will be filled with potholes and populated with monsters. The warning is loud and clear: proceed with caution, and pay attention to the road signs. We have a simple choice: doomsday or utopia. Will we heed the message?
33
remember, but to make a truly intelligent machine, it needs the ability to learnto recognize when it makes a mistake and correct it. Similarly, they need to adapt to changes in their environment. In other words, they need to notice when things that theyve learned no longer apply and be able to correct this error. In order to do all of this, it needs to know its own abilities and limitations so it can learn accordingly. For an unintelligent being (e.g. a robot) to have intelligence, it needs to have goals and desires. It can remember what it has done wrong and work to fix its mistakes, but until it knows what it wants or needs to do, it will not be very useful. It would also need to know a bit about the world around it and be able to make predictions. This could be from common knowledge: for instance if its dark outside but its in the middle of the day, it may need to protect itself from rain; or from observationsif an owner always asks for the same thing at 9:23 AM, the robot would notice this and have it ready beforehand. Of course, in order to be intelligent, it must do all of these things well. If a serial killer calls the house every day for a week when the intelligent computer is still beginning to learn, and if the computer knows that serial killers are not good people, it would very easily assume that the telephone is evil. This can be avoided with a working knowledge of household appliances, but when an unexpected situation arises, it needs to be able to figure out if its observations are correct and worth remembering. The standard litmus test for animated intelligence is known as the Turing Test. Alan B. Turing first proposed it in 1950 and it states that a computer is intelligent if a person cannot tell the difference between it and an actual person.
34
35