Several years ago artificial intelligence was still just a matter of sci-fi movies and theoretical deliberation. Nowadays it is an integral part of our reality. Smart home devices, iOS assistant Siri, or automatic traffic control systems are just a few examples of practical use of AI. Of course, someone might ask: “Can you really call a machine intelligent? Isn’t it just some system of scripts and algorithms, but nothing close to human reasoning and cognition?” To answer this question, we need a proper definition of Artificial Intelligence.
What Do We Mean by “Artificial Intelligence”?
The simplest answer to that question would be: “a branch of computer science focused on developing computers capable of performing tasks normally done by humans”. With that definition, almost every computer program that simulates human intellectual abilities can be described as artificial intelligence. The way it does it is irrelevant.
A good example of such machine is the famous Deep Blue. Designed by IBM, it was the first computer to defeat a chess grandmaster (namely, Garry Kasparov – in 1997). Deep Blue was able to evaluate and choose the most strategic move in a game of chess, but it did it in a different way than human player. That’s why it had a very limited scope of application.
On the other hand, we can think of systems that are really intelligent – in a sense that they can learn, adapt, and remember the previous experience. This abilities will allow them to find a solution even if presented with a new and unfamiliar problem. This type of intelligence is commonly referred to as the “strong AI” (as opposed to the “weak AI”, which refers to systems similar to Deep Blue).
Strong AI vs Weak AI
Weak (or narrow) AI is a type of artificial intelligence designed to perform a specific task. These systems might be very sophisticated and even better than human in their narrow specialization. We have already created computers that are unbeatable in checkers, chess, go, or scrabble. However, they have no real consciousness nor any genuine intelligence. Another example of the narrow AI is Siri, a personal assistant available for iOS users. It uses voice recognition technology and can answer questions or respond to commands, but cannot work outside the predefined range of functions.
Strong (or full) AI is a computer that could not only act like a human being, but also think as one. It is also known as artificial general intelligence (AGI), since it would be able to accomplish any intellectual task possible for human. There are many definitions of requirements that a machine has to meet in order to be recognized as the strong AI. One of the most popular is the Turing test, an idea proposed by British mathematician and computing pioneer Alan Turing. This test was designed to determine whether a machine can be called “intelligent”. One person – a judge, would talk to a several interlocutors. Most of them are human, but one is an artificial chatbot. If the judge cannot tell if he is talking to a human or a machine, then we can say that machine passed the test.
No computer has done it so far. There were many tries, one of the most notable being Eugene, a chatbox created by a group of Russian programmers. Initially, its creators have announced success, but later their statement met with skepticism. American cognitive scientist Gary Marcus described Eugene only as “cleverly-coded piece of software”, but not an equivalent to a system with real, human-like intelligence.
Many other attempts were also unsuccessful. Therefore, the development of AI raises many doubts. Some scientists even claim that creating genuine strong AI may never happen.
Is Strong AI Really Achievable?
With the current state of technological knowledge – It isn’t. There are few barriers that are yet unsurpassable. The computer hardware is still too weak; estimated human brain capacity is around 2.5 petabytes, an amount of data impossible to store and process in real time by any man-made machine. Funds allocated to research of AI are growing, but are still relatively small. It is also extremely difficult to integrate multiple complex elements into one coherent system.
There are also existential and ethical doubts. Many scientists and inventors – such as Elon Musk or Bill Gates – call for extreme caution regarding AI research. When asked about dangers of artificial intelligence, Musk replied: “I’m increasingly inclined to think that there should be some regulatory oversight, maybe at the national and international level, just to make sure that we don’t do something very foolish.”
However, thinking only in terms of laws of physics, artificial general intelligence is possible to create. It is a matter of time, maybe a lot of it, but scientists generally agree that sooner or later technology will allow us to create a machine able to solve complex problems, plan, learn, or communicate in natural language. It will definitely be the greatest invention of mankind – and one to have significant consequences for its future.
Should We Be Afraid of AI?
Vision of robots rebelling against humans and taking over the world is well known from books and movies. These fears are not completely unjustified. We might not be able to predict the behavior of self-aware, learning machines. Technological singularity hypothesis coined by American futurist Raymond Kurzweil states that at some point AI might slip out of human control and start self-improving process which will result in a superintelligence far surpassing human intellectual abilities.
More mundane concerns relate to the economy and the labor market. Researchers predict that even half of human jobs might be replaced by machines. Google self-driving car is one of the examples where a person may soon be redundant. The impact of robot workers on society will be unprecedented. What will all the unemployed humans do then?
Since the creation of AGI is a matter of future, these fears are still in the area of theoretical considerations. However, the important questions have to be asked, and mankind must be prepared for the results of its own creation.
So Are There Really Intelligent Machines?
The answer to the question asked in the first paragraph of this article really depends on our definition of artificial intelligence. Siri, self-driving cars, or characters in video games are all some kind of AI. They are very limited in their doings, but, in a way, they do resemble human beings. If we consider it sufficient, we can describe them as “intelligent” (although the more accurate and commonly used word may be “smart”).
The way to create a genuinely intelligent machine is long and bumpy. It is impossible to estimate a realistic date when it will happen. However, it is no longer a matter of “if”, but rather “when”. Artificial intelligence is already becoming an inherent element of modern world and will only have even greater impact on our lives. What was once a vision of distance future will soon be our present.