Can you imagine having an Artificial Intelligence assistant with the ability to detect when you are in a bad mood and prepare your favorite dish to make you feel better? The news we see every day on television, in traditional newspapers, and on many Internet sites seem to suggest that this type of technology is just around the corner.

These are the four types of Artificial Intelligence that exist so far:

1. Reactive machines

You may have heard of Deep Blue, the IBM-made supercomputer that defeated chess grandmaster and world champion, Garri Kasparov, in 1997. According to Hintze, Deep Blue belongs to the most basic Artificial Intelligence type in existence: reactive machines. This type of AI does not have the ability to store memories or use past experiences to make decisions.

Reactive machines perceive the world directly and act on what they see. Their operation does not require them to create a model of the world. For example, Deep Blue can identify the pieces on a chessboard, make predictions about what his opponent’s possible next moves are, and choose the best responses, but he knows nothing of the past or of the experiences he has had in previous games.

What is relevant about this type of AI is its ability to make the right choice from among millions of possibilities, according to the immediate situation that arises. Both Deep Blue and AlphaGo, the computer created by Google that has already defeated several champions of the Japanese board game Go, are machines that do not have a conception of a world beyond the specific tasks they created. These machines will always behave in the same way when they encounter the same situation. This can ensure that an Artificial Intelligence system is reliable, for example, in the field of autonomous cars, which are expected to be reliable drivers, explains the researcher.

2. Machines with limited memory

Autonomous cars also include this other type of Artificial Intelligence, with which they can take a look at the past. This class of cars has the ability to monitor speed and direction for a specified period. According to Hintze, this data is added to the representation of the world that has been loaded into the computer, which includes the location of traffic lights, traffic signs, or the curves of a road.

But as the name suggests, there is little memory for these types of Artificial Intelligence devices. The data about the cars that pass around them, in the case of autonomous cars, are not stored in a library of information from which they can learn, as is the case with human drivers who learn from the experience they have lived behind the wheel.

3. Machines with a Theory of Mind

People, animals, plants and now, some objects have thoughts and emotions that directly affect their behavior, the study and awareness of this phenomenon is known as Theory of Mind. For Hintze, this is the main characteristic that differentiates the machines that have been built so far from those that will be developed in the future. It is expected that this type of Artificial Intelligence will not only have its own conception of the world in general, but also of precise entities within it, such as the emotions and ideas of which we have spoken previously.

This kind of recognition has been crucial in the formation of human societies. If we don’t understand each other’s intentions and motives, and if we don’t take into account what someone else knows about me or the environment, working together is a much more complicated, if not impossible task, he explained.

In this sense, in order for us to observe machines with Artificial Intelligence walking among us, they must learn to recognize and understand the emotions and sensations of those around them to adapt their behavior to them.

4. Machines with their own conscience

Machines that are capable of building a representation of them will be the last step of the Artificial Intelligence system. Artificial Intelligence researchers must not only know how consciousness works, but we must build machines that have one, Hintze explains.

Those beings with their own conscience know their internal states and therefore are able to predict the feelings of others. For example, we can identify that a person suffers pain when they have fallen because we have felt pain when we fell. According to the professor, we are still far from creating machines that are aware of their own existence, so researchers’ efforts should focus on understanding how memory works, learning, and the ability to make decisions based on past experiences.

For read more articles related to artificial intelligence, click here