What is an "intelligent system" and when is it allowed to talk about "Artificial Intelligence"?

Today nearly everything is sold as being "intelligent". But what are the minimum requirements for being called "intelligent"?

From "intelligent sensors", "intelligent washing machines", "intelligent tooth brushs" to "intelligent materials", nearly everything is being called "intelligent" today. But after looking closer only few are able to fullfil the promises.

For example not the human hand with all its manifold kinematic capabilities and its sophisticated touch senses is intelligent. The functionality of a hand can be estonishing and fascinating but it only becomes an intelligent device by the control of the brain, which knows how to apply all the features in the best way. Hence, intelligence in not located in the sensors and actors. It does not emerge just by the collection of information. It comes from the algorithms and logics, which draw the right conclusion from the sensor information to control the proper actions to act "intelligently".

Under consideration of various existing scientific definitions for intelligence, a technical system at least has to show following properties to be called "intelligent":

  • Detection and interpretation of the situation.
  • Judgement and pondering upon different action alternatives.
  • Adaption to changing environments and new situations.
  • Goal-oriented procedures.

The above definition is derived amongst others from Russel, Norvig: Artificial Intelligence, A Modern Approach, Chapter 2.

Accordingly the pure usage of Artificial Neural Networks or Deep Learning does not necessarily imply Artificial Intelligence, even Artificial Neural Networks feature a lot of beneficial properties for building Artificial Intelligence.

Last update on 2020-11-15 by Andreas Kuhn.

Go back