Artificial intelligence

Definition

Artificial intelligence (AI) is a branch of computer science that deals with the development of systems that exhibit human-like cognitive abilities [1]. This means that AI systems are designed to solve tasks that normally require human intelligence – such as problem solving, learning, planning, perception or language comprehension. An AI system therefore attempts to artificially replicate aspects of human thought or action. Such a system can be a computer program, a robot or even a complex model that runs in a cloud.

A simple definition is: "AI is the ability of an IT system to exhibit 'intelligent' (human-like) behavior." [2]. Depending on the point of view, either the result (a system behaves intelligently) or the functionality (it uses certain algorithms that enable intelligence) is emphasized.

Sub-areas of AI

AI is very broad and encompasses several sub-disciplines:

• Machine learning (ML): systems that learn from sample data. This is a core area of AI today.
• Neural networks & deep learning: approaches inspired by the brain that learn from simple computing units using large networks.
• Natural language processing (NLP): Teaching machines to understand and generate language.

• Computer vision: Understanding images and videos using AI.

• Robotics: AI for controlling physical robots, including perception and navigation.

• Knowledge-based systems and logic: Earlier AI discipline based on explicit knowledge possession and reasoning (e.g. expert systems).

• Planning and optimization: AI methods that plan sequences of actions (e.g. route planning) or solve complex optimization problems.

These sub-areas often overlap. Modern AI systems such as autonomous vehicles combine computer vision (for recognizing the environment), machine learning (for decision-making) and robotics (for vehicle control).

Weak vs. strong AI

In AI, a distinction is made between weak AI and strong AI.

• Narrow AI or weak AI refers to today's systems that master specific tasks but do not possess general intelligence. Examples: A program that only plays chess or a voice assistant that can announce routes - very useful, but clueless outside its narrow field.
• Strong AI would be a system with general intelligence comparable to that of a human being. This long-term goal of AI research – an AI that can solve any intellectual task as well as a human – has not yet been achieved. Virtually all current AI applications fall under weak AI, however advanced they may be.