What is AI?


Definition:- Artificial Intelligence (AI) can be defined as the science of developing human-controlled and operated machines, such as digital computers or robots, that are capable of imitating human intelligence, adapting to new inputs, and performing human-like activities.

Humans or Homo Sapiens often claim to be the most superior species to have ever inhabited the planet Earth, mainly attributing to their “intelligence.” Even the most complicated animal behavior is never regarded as smart, but intelligence is attributed to the simplest of human behavior. For example, when a female digger wasp returns with food to her burrow, she deposits the food on the threshold and checks for intruders prior to carrying her food inside. Sounds like wasps are quite smart, right? However, an experiment performed on these wasps, where the scientist moved the food a few inches away from the burrow’s entrance while the wasp was inside the burrow, reported that the wasps continued to repeat the entire ordeal as often as the food was moved. This experiment revealed that the wasps fail to adapt to evolving conditions and therefore, “intelligence” in the wasps is noticeably absent. Then how do we define “human intelligence?” Psychologists characterize human intelligence as a composite of a variety of skills such as learning from experiences and adapting consequently, understanding abstract ideas, reasoning, problem-solving, linguistic use, and perception.

Thanks to Pop culture, upon hearing the words Artificial Intelligence, most people tend to think of robots coming to life to wreak havoc on human beings. But that is far from reality. The core principle of Artificial Intelligence is the capacity of Artificial Intelligence powered machines to rationalize (think like humans) and take action (mimic human actions) to achieve the targeted objective. To put it simply, Artificial Intelligence is designing a machine that will think and behave like human beings. Artificial Intelligence has three primary objectives, which are learning, reasoning, and perception.

Although the term Artificial Intelligence was coined in 1956, the British pioneer of Computer Sciences, Alan Mathison Turing, carried out extensive work in the field of Artificial Intelligence, in the mid-20th century. Turing created an abstract computing machine in the form of symbols, in 1935, with a scanner and unlimited memory. The scanner was able to move back and forth through the memory, read the available symbols, and write additional memory symbols. A programming instruction dictated the activities of the scanner and was stored in the memory as well. Thus, Turing created a machine with implicit learning capacities that could alter its programming and enhance itself. This principle is commonly regarded as the universal “Turing Machine” and serves as a basis for all advanced computers. Turing asserted that computers are able to learn from their own experience and resolve problems using a guiding principle called “heuristic problem-solving.”

The early Artificial Intelligence studies were concentrated on problem-solving and symbolic techniques, in the 1950s. By the 1960s, the AI research had a major leap of interest from the US Department of Defense, who started working towards training computers to mimic human reasoning. In the 1970s, the Defense Advanced Research Projects Agency (DARPA) has successfully performed its street mapping projects. It may come to you as a surprise that in 2003, long before the existence of the renowned Siri and Alexa, DARPA had effectively manufactured intelligent personal assistant machines. Long story short, this groundbreaking achievement in the field of Artificial Intelligence has set the stage for automation and reasoning observed in modern-day computers.

Here are the primary human characteristics that we strive to imitate in the machines:

  1. Knowledge
    Machines need an abundance of data and information related to the world around us, in order to be able to behave and respond like humans. In order to implement knowledge engineering, Artificial Intelligence-powered machines are required to have undeterred access to data objects, data categories, and data properties as well as the relationship between them that can be managed and stored in the data storage.
  2. Learning
    Of all the various forms of learning that apply to AI, the “trial and error” technique is considered the easiest. For example, a computer chess learning program will try all possible moves until the mate-in-one move is found to win the game. The program then saves the winning move to be used the next time it encounters a similar scenario. This fairly easy to implement learning element is called “rote learning”, which includes memorization of individual objects and processes in a straightforward way. The most difficult part of learning is called “generalization,” which implies the application of prior experiences to the newly encountered similar situations.
  3. Problem Solving
    The systematic process of achieving a predefined objective or solution by looking through a range of viable actions can be characterized as a problem solving the methods for problem-solving can be adjusted for a specific issue or used for a broad range of problems. A general-purpose problem-solving technique frequently used in AI is “means-end analysis,” which includes an incremental decrease in the difference between the initial and final state of the goal. Think of some of the core functions of a robot, back and forth motion, or picking up factors that contribute to an objective being fulfilled.
  4. Reasoning
    The act of reasoning can be defined as the capacity to draw inferences suitable to the scenario in question. The two types of reasoning are called “deductive reasoning” and “inductive reasoning”. In deductive reasoning, if the premise is true then the conclusion is presumed to be true. On the other hand, the conclusion may or may not be true in inductive reasoning, even if the premise is true. While significant success in programming computers to execute deductive reasoning has been attained, the application of “true reasoning” stays out of reach and one of the greatest challenges facing Artificial Intelligence.
  5. Perception
    The process of producing a multidimensional view of an object by means of multiple sensory organs can be defined as perception. A number of variables, such as the viewing angle, the direction, and intensity of the light, and the extent of contrast generated by the object with the surrounding area, can complicate this development of awareness of the environment. With the introduction of self-driving cars and robots that can collect empty soda cans while moving through the facilities, breakthrough developments have been rendered in the field of artificial perception and can be readily observed in our daily lives.

History of Artificial Intelligence

The history of artificial intelligence may sound like a deep and impenetrable subject for individuals who are not well-acquainted in computer science and its related fields.
Regardless of how mysterious and untouchable artificial intelligence may look, when it’s broken down, it becomes easy to understand than you might imagine. So, what is artificial intelligence, otherwise referred to as “AI”? AI is a branch of computer science that concentrates on non-human intelligence or machine-driven intelligence.

AI relies on the concept that human thought functionality can be replicated. Early thinkers pushed the concept of artificial intelligence throughout the 1700s and beyond. During this period, it became more tangible. Philosophers imagined how the idea of human thinking could be artificially computerized and transformed by intelligent machines. The human thinking that generated interest in AI eventually resulted in the invention of the programmable digital computer in the 1940s. This particular discovery finally pushed scientists to proceed with the concept of building an “electronic brain,” or an artificially intelligent being.

Besides that, mathematician Alan Turing had developed a test that ascertained the potential of a machine to emulate human actions to the degree that was indifferent from human actions. From the 1950s, many theorists, logicians, programmers broadened the modern mastery of artificial intelligence as a whole. During this period, intelligence was viewed as a product of “logical” and “symbolic” reasoning. The reasoning was executed by computers using search algorithms. The focus during this time was to replicate human intelligence by solving simple games and proving theorems. Soon, it became obvious that these algorithms could not be used to find solutions to problems such as the movement of a robot in an unknown room. Extensive knowledge of the real world would have been required to avoid a “combinatorial” explosion of the problem to solve.

Come in the 80s; it was pragmatically accepted to restrict the AI scope to specific tasks like the replication of intelligent decision making for the medical diagnosis of particular pathologies. This was the time of “expert systems”, capable of successfully replicating the intelligence of a human specialist in small and well-defined industries. Similarly, it became apparent that some intelligent actions such as recognition of written text, could not be achieved with an algorithm designed with a sequence of instructions previously set. Instead, it was possible to gather numerous examples of the objects to be identified and using algorithms that could master the essential characteristics of these objects. It was the beginning of what we now refer to as “machine learning”.

The computer learning steps could be defined as a mathematical optimization problem and described with probabilistic and statistic models. Some of the learning algorithms that were used to replicate the human brain were referred to as “artificial neural networks”. During the first four decades, AI has moved through moments of Euphoria, followed by times of unfulfilled expectations. In the early 2000s, the increasing emphasis on specific problems and the increase of investments resulted in the first historical accomplishments. In some functions, AI systems attained higher performances than humans.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.