What exactly is Artificial Intelligence?
What does Artificial Intelligence mean?
A human brain consists of about 85 billion nerve cells – a network of nerve cells (neurons) that have connections to their neighboring cells and constantly exchange information. They are the basis for our ability to learn and abstract.
What does this have to do with Artificial Intelligence (AI)? Although the human brain can no longer keep up with the speed of modern computers, with its adaptability and efficient processing of information it (still) beats AI algorithms. The complex functioning of the brain is what scientists are trying to imitate, for example with artificial neural networks (ANN).
In the field of artificial intelligence, a ANN has the task of structuring and grouping information. AI means pattern recognition through human-like perception and decision structures. How strong (more on this later) an artificial neural network is depends on the ability to learn more or less independently from training data.
Even if the subject area appears new to us due to media attention, definitions and projects in this field already existed more than ten years ago:
Source: Russell und Norvig 1995
In the context of artificial intelligence, data scientists and scholars often speak of machine learning. This is one part of the large research area AI. It deals with the development of connections and rules on the basis of a large amount of sample data.
Deep learning means the algorithm learns from examples in which the characteristics are not explicitly specified. Huge amounts of data are analysed, evaluated, logical conclusions are drawn and solutions are selected. Thus, it is a section of machine learning that is characterized by complex solution methods.
Strong or weak AI?
In science and economy a differentiation is made between weak and narrow artificial intelligence. Based on algorithms developed for a specific task, weak artificial intelligence supports people in various activities. It aims at solving clearly defined application problems and does not transfer conclusions to other areas.
This is the case, for example, when it comes to character and image recognition in areas such as voice control, navigation, spam filters or the target-group-specific selection of advertising. One of the best known examples of weak AIs is the famous chess tournament 1997 in which the then world chess champion Garri Kasparov lost a chess competition in the USA against the IBM computer Deep Blue.
Scientists do not yet speak of narrow artificial intelligence in the existing application scenarios. The narrow artificial intelligence should find solutions even for unfamiliar tasks, be able to transfer conclusions and solve problems without human intervention.
A so-called super intelligence is also possible (Strong Artificial Intelligence). In this case, the technology would even exceed the intellectual abilities of humans. However, despite the rapid technological developments in recent years, all previous projects are still in the range of weak AIs.
If you would like to get more involved with the topic, you will find numerous courses and opportunities to develop your own knowledge at providers such as Coursera, Udacity or edX. The use of open-source software can also be helpful for the implementation of AI projects (e.g. the TensorFlow developed by Google).
Are you interested in finding out how you can use artificial intelligence to optimize your processes? Contact our team and boost your business success.