The 7 types of artificial intelligence

The famous John Mcarthy, a prominent American computer scientist who received the Turing Prize in 1971 for his contributions to the In computing, he first coined the term "Artificial Intelligence" (AI) in 1956, during the conference of Dartmouth. This word refers to the use of computers and other technologies for the purpose of simulate intelligent behavior and critical thinking comparable to that of a biological human being.

Human and other animal intelligence involves consciousness and emotionality, while AI today is a conglomeration of primarily objective numbers and guidelines. In any case, an intelligent agent is not one who follows a code perfectly, but one who is capable of perceive their environment and take actions "autonomously" that maximize the chances of success of the task at hand perform.

At the social level, an AI has connotations that human beings expect from an “intelligent” entity, how to reason, perceive, learn and be able to solve problems based on impositions environmental The line between computing and AI is very fragile because, as the processes to “order” a machine to do something are mastered, the autonomy in the “thinking” of the program is limited. This phenomenon is known as the “AI effect”, the peculiarities of which we reserve for another opportunity.

To the general population, AI sounds like an ethereal, fantastic and difficult to understand entity: nothing more far from reality, as intelligent computational models increasingly surround us without giving us account. To put this reality in perspective, today we explain the types of artificial intelligence and their characteristics. Do not miss it.

• Related article: "Theories of human intelligence"

What are the types of artificial intelligence in machines?

An AI is not an anthropomorphic shaped robot serving tea in a domestic environment, although historically the term is associated with such fantastic situations. Without going any further, Siri, the voice assistant of iOS, is considered an artificial intelligence when use, as it is based on natural language processing to meet the needs of the consumer. It is not a robot, but it cannot be refuted that it is an AI model that we all use without realizing it.

If we understand AI as a broad spectrum at a computational level, we will not be surprised to learn that the number of businesses that use it has increased by 270% in the last 4 years. From product recommendations while browsing the internet to the operation of smartphones, artificial intelligence helps us surrounds everywhere: computational intelligence goes hand in hand with social demands, at the research level and even in the field sanitary.

However, an early split is necessary: ​​not all artificial intelligences are the same. We present the types of AI, based on 2 very different parameters: capacity and functionality. Go for it.

1. By capacity

Although those outside the subject consider that we are in the rise of artificial intelligence, human society has only discovered the tip of the iceberg as far as this subject is concerned. You will understand what we mean in the following lines.

1.1. Narrow artificial intelligence (narrow AI)

This variant corresponds to the vast majority of artificial intelligences present on Earth today. This type of AI It is trained to perform a specific and very limited type of activity, so it can fail unpredictably if it tries to act beyond its limitations. Despite the fact that it is a rational entity, its range of action is very narrow, hence its qualifier.

Siri is a perfect example of narrow artificial intelligence, as it operates excellently, but in a very limited range of predefined functions. Other specific cases are programs that play chess, cars that drive themselves and mechanisms that recommend advertising based on our searches.

1.2. General artificial intelligence (general AI)

This kind of artificial intelligence can, theoretically, perform any type of task with the same effectiveness as a biological human being. We speak from a theoretical point of view, since at present the general AI continues in a hypothetical framework, since it has not been developed.

While narrow AI has not been conceived with the idea of ​​performing activities of a cognitive and marked by the "personality" as in the human being, the general AI does aspire to reach this field in some moment. It is not about implementing a framework of action and instructions in the machine itself, but rather simulating the processes inside human brain that allow the computational entity, in theory, to carry out any activity with the same autonomy as a human. Today, more than 40 organizations are studying the field of general AI.

1.3. Super AI

Again, we are faced with a term that is a chimera today. A super AI must be able to perform any activity better than the human being and, in addition, present the ability to think, reason, solve complex questions, apply own judgments, plan based on experience, learn and communicate on their own.

This term is a real challenge in the world of research, since it is still debated whether it is even possible to reach this point at some point in human history. Some authors argue that, since the brain is a mechanical system, it should be possible to simulate it using synthetic materials. However, the great differences and changes in human thinking suggest that reasoning systems based on the very nature of our species with even more complex capacities are both a physical and biological.

• You may be interested in: "Are we rational or emotional beings?"

2. For its functionality

From here, we will go a little faster, as we abandon conjectural grounds and focus on the usefulness of artificial intelligence.

2.1. Reactive machines

The purely reactive machines they are the simplest type of AI that can be conceived. They do not store memories or past experiences in order to implement them in the future, because they simply focus their range of action at a specific time and "try to do their best" with the information available in the now.

2.2. Limited memory

These computational entities are able to store past experiences or data for a short and limited period of time. An excellent example of this type of AI is artificial cars, since they “remember” recent data in order to carry out their task as best as possible, such as speed limit, route to follow, safe distance between 2 vehicles and other basic parameters.

2.3. Theory of mind AI

This type of AI should be able to understand human emotions, social constructs, beliefs and other parameters to be able to interact with us as 2 people would. We speak in the conditional, because the machines that apply the theory of mind have not yet been designed.

• You may be interested in: "Theory of Mind: what is it and what does it tell us about ourselves?"

2.4. Self-awareness machines

Self-awareness is one of the first and most ambitious goals in computational research right now. A self-aware machine must not only be able to store past data, but create its own judgment based on them and act according to the autonomous entity it deems appropriate., thus adding such complex terms to the equation as feelings and values.

Resume

As you can see, the only AI available today is the narrow type, either in the form of a reactive machine or limited memory. In any case, with these insights we have not wanted at any time to detract from the historical milestone that has artificial intelligences today. An AI is programmed to perform a task, yes, but you must not forget that it does so in the most effective way possible and responding to environmental variations with expertise.

Defining the limit of programming and AI is a less complex debate, because the more you know, the easier it is to program a machine to do exactly what we want. Of course, the future of artificial intelligence lies in general AI and in computational entities capable of developing self-awareness. Only time will tell if the limit is biology.

Bibliographic references:

• Dick, S. (2019). Artificial intelligence.
• Došilović, F. K., Brčić, M., & Hlupić, N. (2018, May). Explainable artificial intelligence: A survey. In 2018 41st International convention on information and communication technology, electronics and microelectronics (MIPRO) (pp. 0210-0215). IEEE.
• Hosny, A., Parmar, C., Quackenbush, J., Schwartz, L. H., & Aerts, H. J. (2018). Artificial intelligence in radiology. Nature Reviews Cancer, 18 (8), 500-510.
• Lu, H., Li, Y., Chen, M., Kim, H., & Serikawa, S. (2018). Brain intelligence: go beyond artificial intelligence. Mobile Networks and Applications, 23 (2), 368-375.
• McCarthy, J. (1998). What is artificial intelligence ?.
• Types of artificial Intelligence, Javapoint.com. Picked up on March 18 in https://www.javatpoint.com/types-of-artificial-intelligence
• Yu, K. H., Beam, A. L., & Kohane, I. S. (2018). Artificial intelligence in healthcare. Nature biomedical engineering, 2 (10): pp. 719 - 731.