Artificial intelligence made simple

Published on: October 31, 2018 / Update from: April 8, 2019 - Author: Konrad Wolfenstein

Artificial intelligence simply explained. Maintaining an overview in the mass, e.g. Big Data? This is only possible if you follow certain patterns or let yourself be guided.

A self-experiment: You have a certain image in your head. Today it should be a red cabinet with white handles. What are you doing?

You enter “red cabinet, white handles” in Google search.

Yield? Modest.

2nd attempt: You enter “red cabinet, white handles” in Google search.

The result is already better, but could certainly be even better.

The first step into programming is taken with the Google search. The collection of search queries and the conversion of them into algorithms and codes form the neural network.

Machine learning, as shown in the top graphic, is therefore not a thing for quick implementation. A lot of time and work goes into it. This also explains the corresponding development costs. But if you consider that the AI ​​has no vacation, no pension or other natural losses, things look completely different.

But will the red cabinet with white handles still be relevant tomorrow? Does it still fit the lifestyle? Tastes change. This is exactly where deep learning comes into play. To stay with our example: As the search continues, the AI ​​learns and recognizes how your search behavior has changed based on the other topics that interest you and independently develops new algorithms to “anticipate” that you will have a green cupboard in a year with blue handles could be of interest for the kitchen.

Terrible? For some this is frightening. But it's actually not. The fear of the unknown plays tricks on us. If we asked a group of people what might interest you on TV tomorrow, you would get a variety of answers. Not uniform. Now, how do you decide which proposal you would accept? Is it the professional contribution or the attractive appearance of the person in question?

It's the same with AI. The statement depends on how weak or strong the neural network has been “programmed”. It's about pattern analysis to help us make a good decision. Not to control us. Because if we don't manage to analyze patterns in big data, we'll go under without mercy. And that is the real horror scenario.