What Is Artificial Intelligence? What Does It Do?

What is Artificial Intelligence? What Does It Do?

AI (pronunciation: EY-AY) or Turkish equivalent Artificial Intelligence ; It is a simulation of processes that predict human intelligence, especially by computer systems and machines. These processes include learning (rules for acquiring and using information), reasoning (using rules to achieve approximate or precise results), and self-correcting. Artificial intelligence; special applications, expert systems, speech recognition and machine vision.

AI was shared for the first time at the Dartmouth Conference in 1956 by American computer scientist John McCarthy. Artificial intelligence is an umbrella term that covers everything from robotic process automation to real robotics today. It has increased in popularity recently due to big data or an increase in the speed, size and diversity of businesses. Artificial intelligence can fulfill tasks such as using patterns in data more efficiently than people or providing businesses to have more information than the data they have.

Artificial Intelligence Types:

Artificial intelligence can be categorized in various ways, basically two examples can be given.


First, we can classify AI systems as Weak AI and strong AI. Weak AI is also called Narrow AI. It is an AI system designed and trained for a particular task. Virtual personal assistants like Apple's Siri are the Weak AI type.


Powerful AI, also known as artificial general intelligence, is an AI system with generalized human cognitive abilities, so it has enough intelligence to find a solution compared to an unknown task. The first example of this, the Turing Test, is a method developed by the mathematician Alan Turing in 1950. Although the method is controversial, it is a method used to determine how a computer can think like a human.


The second example came from Arend Hintze, an assistant professor of integrated biology and computer science and engineers at Michigan State University. It divides AI into four different categories, from existing AI systems to perception systems that are not yet available. The categories are:

Type 1 Reactive Machines:

 Analyzes possible movements - self and opponent - and selects the most strategic movement. As an example, we can give the IBM chess program Deep Blue, which defeated Garry Kasparov in 1990. Deep Blue can detect and predict pieces on the chessboard, but has no memory and cannot use past experiences for future events. Deep Blue and Google's AlphaGO were designed for narrow purposes and could not be easily applied to any other situation.

Type 2 Limited Memory:

  These AI systems can use their past experience to learn about future decisions. Some of the decision-making functions in autonomous vehicles are designed in this way. Current practices show the near future use of vehicles that can follow lane lines. Data in AIs whose memory is limited is not permanently stored.

Type 3 Mind Theory:

  This is a psychology term. This expresses the understanding that influences the decisions made by others or that they have their own beliefs, desires and intentions. This AI type does not yet exist. It is possible that we will see this application in social media tools or smart phone applications in the future, if not now, and we will be aware of it.

Type 4 Self Awareness: 

In  this category, AI systems have a unique sense and consciousness. Machines with self-awareness understand their current feelings and positions and can use the information to influence and direct others. This AI type is not yet available.

Artificial Intelligence Technology Examples:

Automation is the process of automatically performing a system or process function. Robotic process automation can be programmed to perform high data, repeatable tasks gradually done by humans. RPA ( Robotic Process Automation ) differs from IT (Information Technology) automation in that it can adapt to changing conditions.
Machine learning is the process of a computer taking action from programming. There are three categories of machine learning algorithms.

Deep Learning : 

In simple terms, its prediction can be defined as analytical automation.
Supervised Learning ; data sets are tagged here so that patterns can be detected and used to tag new data sets.

Unsupervised Learning:

 It can be sorted by similarities or differences in which data sets are not tagged. Support learning where data sets are not tagged, but feedback is given to the AI ​​system after performing one or more actions.
Machine vision is the way computers see. Machine vision captures and analyzes visual information by converting images of a camera from analog to digital and using digital signal processing. It is often compared to the human eye, but machine vision is not biologically dependent and can be programmed to be seen, for example, between walls. It is used in many areas from signature identification to medical image analysis. Computer vision focusing on machine-based image processing often coincides with machine vision.
Natural language processing (NLP) is the processing of human language and computer language through a computer program. One of the oldest and best-known examples of NLP is the spam detection method that looks at the subject line and text of the email and decides whether it is trivial. NLP tasks include text translation, emotion analysis, and speech recognition.
Model recognition is a branch of machine learning focused on determining models of data.
Robotics is an engineering field that focuses on the design and manufacture of robots. Robots are often used to perform tasks that are difficult for people to perform or to perform. They are used by NASA to transport large objects on assembly lines or in space for automobile production. More recently, researchers are using machine learning to build robots that can interact in social environments.

AI Applications:

In health : the greatest expectation on this issue is to improve patient outcomes and reduce costs. Companies apply the “machine learning” technique in artificial intelligence to diagnose better and faster than humans. One of the best known health technologies is  IBM Watson . Understands natural language and can answer questions. The system processes patient data and other available data sources to form a hypothesis, which is then presented with a confidence scoring scheme. Other AI apps include chatbots, a computer program that performs virtual health assistants that help patients during online calls and keep track of appointments that assist customers or provide basic medical feedback during the billing process.

In Business :

 Robotic process automation is applied in multi-repeat tasks performed by humans. Machine learning algorithms are integrated into analytics and CRM platforms to reveal information about serving customers better. Chatbots are included on their website to serve customers instantly.

In training :

 AI can automate rating by giving trainers more time. It evaluates students and can assist the student to work at his or her own level and support the needs appropriately. It can provide support to keep track of teachers, students and other staff at the institution.

In finance:

 AI applied to personal finance applications such as Mint or TURBO TAX provides great convenience in the transactions of financial institutions. Such applications may collect personal data and provide financial advice. Today, such software is used by many companies trading on Wall Street.

In law : 

The issue of whether the texts in the documents are contained in the law is a study that takes quite some time by lawyers and some laws are likely to be forgotten. AI automates this process to support better use of time and accuracy.

In production :

 Including robots in the workflow will increase its importance now and in the future.