In this post, I would like to discuss what artificial intelligence (AI) is and how it can be categorized. Everyone surely has an idea of what artificial intelligence is and what it can do. It’s not uncommon to think of the Terminator or Sonny from I, Robot, for example. Those who were lucky enough to grow up with Star Trek will probably immediately think of Data. But even today, this is still science fiction and falls into the field of strong AI. More on that later.
First, we should clarify the question of what artificial intelligence is. My attempt to put it simply: Artificial intelligence is the attempt to teach a computer or an application how to learn and think like a human being. The aim is to give this application intelligence that it can use to find answers to problems of all kinds and solve them independently.
Weak AI
Weak AI is basically an algorithm. In many cases, it is admittedly a very complex algorithm, but weak AI has no consciousness of its own and cannot demonstrate understanding. It is often trained and used to recognize patterns or search through very large amounts of data. Weak AI performs complex, repetitive tasks that can be precisely specified.
In our everyday lives, we encounter weak AI in text and image recognition (for example traffic sign recognition in cars), speech recognition (chatbots, Alexa, Siri, etc.), navigation systems, advertising and product recommendations (Amazon, Google, Netflix, etc.), predictive maintenance, and so on and so forth. Almost every AI you can imagine that is used today (outside of the cinema screen and science fiction) is weak AI.
Strong AI
Strong AI, also known as superintelligence or general purpose AI (GPAI), for example, is still science fiction from today’s perspective. A strong AI is, for example, a robot that can solve problems of a general nature and is not limited to any specific area of application. It acts independently, proactively, and flexibly, similar to a human being, and can even be creative. As described at the beginning, Data, the Terminator, and Sonny are examples of such systems.
Because this is still science fiction, the list of examples is soberingly short. In another article, we will take a closer look at for example or Transformer.
Expert systems
Expert systems are applications developed to solve complex problems in a defined area. The success of an expert system depends largely on the knowledge database and its quality, and ultimately reflects the knowledge of the expert or experts who created that database.
Deep Blue Source
Probably one of the best-known expert systems is the chess computer Deep Blue, which defeated the then world chess champion Kasparov in 1997. This system was further developed and became famous in 2011 when Watson won the quiz show Jeopardy. Other expert systems are used, for example, in the evaluation of X-rays, analysis of chemical compound structures, geological exploration (reconnaissance, earthquake prediction, etc.) and many other areas.