Artificial Intelligence

Artificial Intelligence (AI) is an aspect of computer science and engineering that deals with intelligent behaviour, learning and adaptation in machines. Most of the research carried out into AI is concerned with producing machines to automate tasks that require intelligent behaviour. Examples of such work would include such areas as handwriting, speech and facial recognition; the ability to answer diagnostic and consumer questions and the ability to control, plan and schedule certain events. The whole area of AI has become a major part of the engineering discipline with a clear aim of seeking to provide solutions to real life problems.

AI has many applications including game playing, speech recognition, understanding natural language and computer vision. Click here for more details.

Click here for more FAQs concerning the basics of AI.

Strong AI and Weak AI

In the philosophy of AI, there is what is known as Strong AI and Weak AI. Strong AI proposes that AI will eventually rival human intelligence. With developments in technology and computers becoming more and more rapid, it is believed that machines can become self-aware and be able to reason and solve problems independently.

On the other hand, Weak AI proposes that machines will never be able to rival human intelligence. Technological developments may well allow computers to be able to solve specific problems or perform reasoning tasks that do not include the full range of human cognitive abilities. However, it is believed that they will not match the abilities of a human and not be able to reach full self-awareness. At best, machines will be able to perform highly intelligent tasks with specific problem solving capabilities.

Two schools of thought have emerged concerning AI. These are as follows:

• Conventional AI
• Computational Intelligence (CI), which is also sometimes referred to as Synthetic Intelligence.

Conventional AI

Under the heading of Conventional AI come methods that are now recognised and classified as machine learning, involving formalism and statistical analysis. Other terms that are used interchangeably with this form of AI are symbolic AI, logical AI, neat AI and Good Old Fashioned Artificial Intelligence (GOFAI).

The main methods that are involved with this form of AI are as follows:

• Expert systems: The process whereby reasoning capabilities are applied in order to reach a conclusion. Such a system is able to process large amounts of known information in order to provide a set of conclusions.
• Case based reasoning
• Bayesian networks
• Behaviour based AI: a modular method of building AI systems by hand.

Computational Intelligence (CI)

This form of AI involves repetitive development or learning. Learning is guided by experience and is therefore based on empirical data, often associated with non-symbolic AI, scruffy AI and soft computing.

The main methods that are involved with this form of AI are as follows:

• Neural networks: systems with very strong pattern recognition capabilities.
• Fuzzy systems: Used in modern industrial and consumer product control systems, such systems provide techniques for achieving reasoning under uncertain conditions.
• Evolutionary computation: This form makes use of biologically inspired concepts such as populations, mutation and survival of the fittest in order to help generate better solutions to the problems.

Robotics

Any form of electro-mechanical or biomechanical device or group of devices that are able to perform an autonomous or pre programmed task, are known as robots. A telerobot is a robot that can act under the direction of a human being, such as the robotic arm on board a space shuttle. Alternatively, a robot may work autonomously under the control of a programmed computer.

Robotics is an area of A.I. concerning the use and practical application of robots. It is also understood to be the term given to the application of robots in order for them to perform manual tasks such as on an assembly line.

At the heart of robotics seems to be a desire to make robots that can do activities that are dangerous, boring, repetitious or indeed impossible for man to carry out. Furthermore, through robotics, increased understanding can be gained concerning how humans behave and work.

Nanorobotics is the technology of creating machines or robots at or below the size of 1-10 micrometer. More specifically, nanorobotics refers to the still largely theoretical nanotechnology engineering discipline of designing and building nanorobots. Nanorobots are typically devices ranging in size from 0.1-10 micrometres and constructed of nanoscale or molecular components.

Within the medical industry, nanorobotics are increasingly being used in minimally invasive procedures. They are being deployed to perform highly delicate, accurate surgery and also used in partnership with a surgeon by way of remote control, to perform surgery.

Modern robotics could roughly be categorised into two main groups:

• Those used in the arenas of entertainment and performing routine tasks.

One primary example of this is found in the car production business where robots are use in the production of specific products, particularly in the area of painting, welding and assembling of cars.

• Those used in the security and medical industry.

Examples of this are found in areas of work such as bomb disposal; work in space, underwater, cleaning up of toxic waste. The iRobot Packbot and the Foster-Miller TALON have been used in Iraq and Afghanistan by the US Military in defusing roadside bombs and other forms of explosive disposal.

As early as the 17th century, Rene Descartes first envisioned the bodies of animals as highly complex creations but nonetheless reducible to machines. This heralded the formulation of his mechanistic theory, which is also referred to as the “clockwork paradigm”. In 1623 Wilhelm Schickard created the first digital calculating machine,which was then followed by inventions by Blaise Pascal (1643) and Gottfried Wilhelm von Leibniz in 1671 who also invented the binary system.

The Turing Test

Many developments were consequently made in the realm of logic and theorem understanding. However it was in the 1950s where active advancements were made in the area of AI. In 1950, the “Turing Test” was pioneered in Alan Turing as a way of administering a test of intelligent behaviour ¹. It was his view that it AI was probably best researched through the programming of computers as opposed to building machines. His article, Computing Machinery and Intelligence discussed the conditions necessary in order to consider a machine to be intelligent. If a machine could successfully pretend to be a human in interactions with a knowledgeable observer then the machine could be considered to be intelligent. Such a test would not satisfy all philosophers but nevertheless it was a worthy contribution to the start of the debate. The test involved two people (A) and (B) sitting in a room while an interrogator (C) sat outside. Person A would try and fool the interrogator concerning their gender, whilst person B would try to assist the interrogator in their identification. Turing’s test went on to suggest a machine take the place of person A. If the machine was able to consistently fool the human interrogator, the machine would be likely to be considered intelligent.

The disadvantage of the Turing test is that it is one sided. A machine that passed such a test could quite rightly be considered intelligent. However, a machine could still be considered human even with a limited understanding (or none at all) of human and human behaviour. Subsequently, in 1951 a draughts playing game written by Christopher Strachey and a chess playing game written by Dietrich Prinz were the first to run a Ferranti Mark I machine of the University of Manchester. These were the very first AI programs.

It was John McCarthy in 1956 who first coined the term “artificial intelligence” at the first conference convened to discuss and explore such a topic. As a result, the birth of AI is considered to be July 1956 at the Dartmouth conference where many people including John McCarthy and Joseph Weizenbaum (who built ELIZA a chatterbot implementing Rogerian psychotherapy) gathered and exchanged ideas.

Furthermore, the work of John von Neumann also contributed greatly to the work and advancement of AI through his development of the game theory.

Various achievements took place during the 1960s and 1970s, with the first successful knowledge based program in mathematics, developments in rule-based systems for knowledge representation and inference in medical diagnosis.

The 1980s saw the first robot cars with one such car driving more than 1000 miles in traffic at up to 110mph in 1995. A chess playing computer, Deep Blue, beat Garry Kasparov in a famous six-game match in 1997. Furthermore, the US governments’ entire investment in AI research since the 1950s was repaid when it decided to implement AI methods for scheduling units in the first Persian Gulf War, according to the Defense Advanced Research Projects Agency (DARPA).

The 1990s and 2000s has seen AI influenced by probability theory and statistics along with increasing interest in big game theory applied to AI decision-making. Following the September 11th terrorist attacks, interest has been renewed in threat-detection AI systems.

For further information on the history of AI, click here.

Robotics

The thinking behind the concept of artificial people and robots can be traced all the way back to the realm of ancient legends. In the legend of Cadmus, it is believed he sowed dragon teeth which turned into soldiers. In the myth of Pygmalion, the statue of Galatea came to life. In Greek mythology, the deformed god of metalwork (Vulcan or Hephaestus) created mechanical servants, ranging from intelligent, golden handmaidens to variations of household items which were able to move about indepedently.This trend continues with various references to articifical “creatures” appearing in medieval, Jewish and Norse traditions.

The actual word, robot, was first introduced by Karel Capek, the Czech writer, in his play R.U.R, (Rossum’s Universal Robots) in 1920. However, before this the verb, robotovat meaning “to work” or “to slave” coming from the noun robota, had been used since the early 10th century in the Czech and Slovak languages.

From a historical perspective, one of the first recorded designs of a humanoid robot was made by Leonardo da Vinci in 1495. Detailed drawings of a mechanical knight, which was able to sit up, wave its arms and move its head and jaw, were found in the 1950s when some of Da Vinci’s notebooks were rediscovered. There is uncertainity as to whether or not Leonardo attempted to build the robot, however it has been suggested that it seems fairly likely his designs bare a striking connection with his anatomical research recorded in the Vitruvian Man. Moreover, in 1738, Jacques de Vaucanson created a mechanical duck which was able to eat grain, flap its wings and excrete.

However, many tend to consider the creation of Nikola Tesla as the first robot in the modern sense of the word. Tesla created a teleoperated boat which was demonstrated at an 1898 exhibition in Madiso Square Garden. He had hoped to develop this “wireless torpedo” into an automated weapon system for the US Navy, in much the same way as he proposed to develop other remotely operated devices, systems and vehicles into fully working prototypes which would advance modern battlefield technology. However, these were only ever proposals and Tesla never actually got to the stage of building them. Interestingly, it is reported that Tesla once said that remote controlled machines were merely precursors of “machines possessed of their own intelligence”.

Elektro, a humanoid robot, was created in the 1930s by Westinghouse. Exhibited at the 1939 and 1940 World’s Fairs the creations attracted much attention. It was not until 1948 at Bristol University, England however that W. Grey Walter created the first electronic autonomous robots.

Nanorobotics are being considered as a way to help ‘find’ and ‘remove’ problems in the body. For example, scientists envisage microscopic robots that could help clean plaque from our arteries and tartar from teeth².

Even eyesight could be corrected by way of nanotechnology and nanorobotics. An electronic “rubber band” placed around the eye to adjust the eye when needed, by way of a switch on the side of the head in much the same way spectacles are worn currently³. This technology is currently being tested and developed.

Furthermore, the work being spearheaded by the Center for Biologic Nanotechnology is also of interest. The Center’s work on Smart Anti-Cancer Therapeutics is based upon how to target cancer cells whilst sparing healthy cells⁴.

So far the Center has been successful in producing four out of the five working prototypes that will make up the completed device. Work is currently being carried out in order create the one device that will perform all of the following tasks:

• Cancer cell recognition
• Diagnosis of cancer cause
• Drug delivery
• Reported tumour location
• Reporting cancer cell death

See the information guide on nanotechnology for further information and examples.

Recent developments in robotics

• Aibo Sony robot dog
  A dog that learns by interaction. Essentially a robot-pet. Sony will cease production of these ‘dogs’ as of March 2006.
  
• Pearl – the nurse bot – personal robotic assistant
  Designed as a help to the elderly, who live on their own, the nurse bot helps those suffering from mild cognitive disabilities and or common physical ailments.
  
• Tribolite vacuum cleaner
  The Tribolite vacuum cleaner uses a sonar system to navigate along with four motors in order to take care of the cleaning all by itself. In the same way a bat ‘sees’, the Tribolite is able to navigate around obstacles in its way.
  
• IRobot PackBot Explorer
  PackBot Explorer allows soldiers to stay at a safe standoff distance while the robot relays back real-time video, sounds and sensor readings. Buildings, bunkers, caves, tunnels, sewers, collapsed structures and other areas that are dangerous or inaccessible to humans can be remotely searched with PackBot Explorer to make sure there are no surprises – like booby traps, mines, weapons caches or enemy soldiers.
  
• Kismet-the social robot
  A robotic head that can interact with humans in a human-like way through various facial expressions, head positions and tones of voices.

The future of robots

Various experts have made their predictions concerning the future development of robots. Click here to read what some of them have to predict.

Recent news stories concerning robots include the following:

• Victor A. Patton, January 2, 2007, Merced Sun Star, Giving robots human qualities
• BBC News Online, January 2, 2007, Mars rovers are taught new tricks
• Technology Review, Wednesday, January 03, 2007, Hearing machines

Click here to view more news stories concerning AI.

As the development of AI develops and the use of robotics are increasingly implemented in every day life, stories and instances previously considered to be in the realm of science fiction are becoming reality. With this comes the need for us to consider and confront important ethical and social questions.

a) Robots vs Humanoids?

At present, humanlike robots are used today to promote and market certain products, causing us to invest in them and trust them more and more with everyday tasks. However, one of the immediate concerns is not that robots will necessary take over and lead us to robot anarchy, but rather they will become more and more life like capable of selling us anything, both good and bad, true and false⁵.

For example, Qrio, Sony’s success story in the world of robotics, is human shaped, and a self propelled puppet that can walk, talk, pinch and take pictures and yet, as some have pointed out, it has no more ethics than a tin can. Isaac Asimov wrote in his classic, I, Robot, in 1950 his dream of robots sitting down together and discussing the ethical implications of their actions. However, it is clear that Qrio will never reason over his actions, nor care for nothing or know anything. Bruce Sterling, writing in Wired magazine, therefore concludes that in light of this, Qrio and its ‘siblings’ to come could quite easily move from being a piece of technological ‘fun’ to be machines which if improperly programmed could shoot down school children walking to school, set fire to someone’s home, slit another person’s throat or move into a shopping centre where upon it detonates itself whilst shouting the cause of an extreme political organisation⁶.

In 1981 the first person to be killed by a robot occurred. 37 year old Kenji Urada, a Japanese factory worker, climbed over a safety fence in order to carry out some maintenance work on a robot. In his desire to get the work completed as soon as possible, he failed to switch the robot off properly. As a result, the robot’s powerful hydraulic arm continued to work, unaware of Urada’s presence. Consequently, the robot accidentally pushed the man into the grinding machine⁷.

All concerns seem to stem from the main theme that should the ability and intelligence of robots exceed that of humans, they could in turn develop a conscience and a motivation to take over or destroy humans.

b) Merging of robot and human

Academics such as Professor Jordan Pollack quite easily envisage a time when humans and computers will be connected via direct neural interfaces, leading to technology that significantly blurs the distinction between telepathy and telekinesis⁸.

Bill Joy, cofounder and chief technologist for Sun Microsystems, wrote a very powerful article entitled “Why The Future Doesn’t Need Us” in the techie magazine, Wired⁹. Much of what he wrote echoed statements and concerns that had previously been raised by environmentalists and activists but what made it all the more powerful was the fact that Joy, one of the gurus of the technological age, was voicing these concerns as well. He begins his article by contrasting the threats of the 20th and 21st century. In his opinion, in the 20th century the top three major threats to human existence were chemical, biological and nuclear warfare. On the other hand, in the 21st century, he argues that the threats to humanity could well become more sophisticated: genetics, artificial intelligence and nanotechnology. Joy concludes that these technologies will probably lead to one of two outcomes. Either by some terrible mistake these new technologies will destroy humankind or they will destroy us in some alternative way, by way of blending human and machine. In doing so, human nature will be superseded by a form of ‘machine nature’ that draws upon a combination of these new technologies. Therefore, by way of either a terrible and significant disaster or through a series of bad choices and decisions, the human nature will be eradicated.

Certainly the work of Professor Kevin Warwick is rapidly pushing the boundaries of human and computer (Click here to download MP3 file of Kevin Warwick speaking at a previous BioCentre conference). Warwick openly expresses his desire to become one with his computer, writing in ‘Wired’ magazine, February 2000¹⁰. By placing an implant in his arm he has sought to pick up nerve impulses from his body and transmit it to a computer. From here the signal can be saved and manipulated and sent back to the implant, the consequences of which are huge. Noting when he is happy, Warwick suggests that the ‘happy’ signal could be recorded and when his mood changes, the signal could be played back to the implant and the results noted. He has already taken this a stage further by having implants placed in two people (himself and his wife) at the same time with movement and emotion signals being sent from one person to the other, via the Internet.

Such technology could be used to treat patients whose central nervous systems have been damaged or affected by disease such as multiple sclerosis or those affected by a stroke. A team from Emory University in Atlanta have implanted a transmitting device into the brain of a stroke patient. Once the motor neurons were linked to silicon, the patient was able to move a cursor on a computer screen just by thinking about it¹¹.

Clearly, alongside issues concerning AI machines, there is the steady development of connections between AI machines and human beings. The word cyborg which perhaps has for many been kept within the domain of science fiction, clearly describes the result of such connections. A cyborg is a cybernetic organism or cyb-org. The concept of linking the human brain with a machine in order to alter the way in which it functions, surely presents some very disturbing consequences. However, this really is no longer an elaborate sci-fi idea destined only to be read about in works of fiction. Experiments of this kind are taking place now, seeking to shape the future.

c) Brain augmentation

Research carried out in laboratories in the USA has led to the creation of a remote controlled rat. The direction of this robot rat can determined by a human using a transmitter standing up to only 547 yards away. It is recognised that almost anything done to a rat can also be done to a human. Therefore, if it is now possible for a remote controlled rat to be created it can surely be possible for a remote controlled human, which in turn leads to a frightening world of violated human dignity¹².

d) Physical augmentation

Whilst advances in robotics may bring about many helpful solutions to life’s problems, such developments may actually cause us to exceed necessary frontiers in technological advances. For example, in Japan which has a rapidly growing elderly population and a clear shortage of carers for such a portion of the population, Japanese roboticists have stepped into the breach and have set about creating a future of walking wheelchairs and mobile arms that manipulate and fetch. Whilst initially this may sound a wonderful solution, it could lead to an ethical hell. The peripherals may be clever and ahead of their day but the human beings remains old, weak, vulnerable, pitifully limited and possibly senile¹³.

e) Robotics & pollutions

Pollack also argues that when cars were first invented no one imagined the consequences of the byproducts, namely carbon monoxide, generated by car usage expelled into the atmosphere. With the increases in robots, Pollack poses the question as to what extent will pollution levels rise due to the increased use of robots, which are often powered by internal combustion engines¹⁴.

Moreover, it has been suggested that robots could well be ‘fed’ biological matter, either food or waste items in order for them to gain energy. If such an approach were to be adopted how would humans feel having to compete for resources with robots?

f) Regulation

A telerobot is a robot which can be controlled by a human using a PC and devices such as a joystick or set of gloves. With the availability of such technology, some have argued that it would certainly present an attractive proposition to employers of replacing an onsite regulator with a handler, based in an overseas call centre. Outsourcing jobs would effectively be taken a step further to mean importing brain power over the internet in order to manage machinery in factories, to teach in schools or to clean homes. The possible financial savings could be well worth considering. However, how would such a situation be regulated? Would local labour laws by enforced on those who would telecommute from a foreign country, such as India?¹⁵

g) Robots & war

As the field of robotics develops, there comes the need to clearly differentiate between autonomous robot weapons and remote control armaments, which are unmanned telerobots supervised by humans. The key ethical question that emerges from this state of affairs is who pulls the trigger?¹⁶ Furthermore, who could be held to account for an unmanned war crime and are we in fact blundering into mechanised killing fields we would have never built by choice?

h) Patents