The 7 types of artificial intelligence

A summary of the most prominent types of artificial intelligence, and their characteristics.

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

Human and other animal intelligence involves consciousness and emotionality, whereas AI today is a conglomeration of primarily objective numbers and guidelines. In any case, an intelligent agent is not one that follows a code perfectly, but one that is able to perceive its environment and take actions "autonomously" that maximize the probability of success of the task to be performed.

On a social level, an AI has connotations that humans expect from an "intelligent" entity, such as reasoning, perceiving, learning and being able to solve problems based on environmental impositions. The line between computation and AI is very fragile because, as processes are mastered to "command" a machine to do something, the autonomy in the "thinking" of the program is limited. This phenomenon is known as the "AI effect", the specifics of which we reserve for another opportunity.

To the general population, AI sounds like an ethereal, fantastic and difficult to understand entity: nothing could be further from the truth, as intelligent computational models increasingly surround us without us even realizing it. To put this reality into perspective, today we explain to you the types of artificial intelligence and their characteristics. Do not miss it.

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.. For example, Siri, the iOS voice assistant, is considered to be an artificial intelligence as it relies 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 using it has increased by 270% in the last 4 years. From product recommendations while browsing the Internet to the operation of smartphones, artificial intelligence is all around us: computational intelligence goes hand in hand with social demands, at the research level and even in healthcare.

However, an early distinction must be made: not all artificial intelligences are the same. We present you the types of AI, based on 2 very different parameters: capability and functionality. Let's get down to it.

1. By capability

Although outsiders consider that we are at the peak of artificial intelligence, human society has only discovered the tip of the iceberg as far as this topic 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 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.. Although it is a rational entity, its range of action is very narrow, hence its label.

Siri is a perfect example of narrow artificial intelligence, as it operates excellently, but in a very limited range of predefined functions. Other concrete 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 type of artificial intelligence can, in theory, perform any type of task as effectively as a biological human being.. We speak from a theoretical point of view, because at present general AI is still 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 nature and marked by "personality" as in humans, general AI does aspire to reach this field at some point. The idea is not to implement a framework of actions and instructions in the machine itself, but to simulate within it the human brain processes that allow the computational entity, in theory, to perform any activity with the same autonomy as a human would. Today, more than 40 organizations are studying the field of general AI.

1.3. Super AI

Again, this is a term that is still a pipe dream today. A super AI must be able to perform any activity better than a human being and, in addition, have the ability to think, reason, and use reasoning, the ability to think, reason, solve complex questions, apply its own judgments, plan based on experience, learn and communicate on its own..

This term is a real challenge in the world of research, as 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 thought suggest that reasoning systems based on the very nature of our species with even more complex capabilities are both a physical and biological impossibility.

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

2. By their functionality.

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

2.1. Reactive machines

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

2.2. Limited memory

These computational entities are capable of storing past experiences or data for a short and limited period of time.. An excellent example of this type of AI are artificial cars, as they "remember" recent data in order to perform their task as well as possible, such as the speed limit, the route to follow, the 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 do.. We speak conditionally, because the machines that apply the theory of mind have not yet been designed.

Self-aware machines (self-awareness) 2.4.

Self-awareness is one of the first and most ambitious goals of computational research at the moment. A self-aware machine must not only be able to store past data, but create its own judgment based on it and act as the autonomous entity sees fit.Thus adding such complex terms to the equation as feelings and values.

Summary

As you have seen, the only AI available today is of the narrow type, either in the form of a reactive or memory-limited machine. In any case, with these appreciations we have not wanted to detract at any time from the historical milestone of having artificial intelligences today. An AI is programmed to do a job, yes, but you must not forget that it does it as effectively as possible and responds to environmental variations with expertise.

Defining the limit of programming and AI is a complex debate to say the least, because the more we know, the easier it is to program a machine to do exactly what we want it to do. 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.

Bibliographical 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. Recogido a 18 de marzo en 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.