Defining what exactly artificial intelligence is is a daunting task. According to De Mauro’s dictionary, it is “the set of studies and techniques that tend to the realization of machines, especially electronic calculators, able to solve problems and reproduce activities typical of human intelligence”. There is not yet, however, a universally accepted definition because AI (from English Artificial Intelligence) is an extremely recent sector and in strong evolution. Bellman (1978), for example, considers it “the automation of activities that we associate with human thought” as “decision making, automatic problem solving, learning … “, while Knight (1991) would be the “study of mental faculties through the use of computational models”.
O Stubblefield (1993), which identifies it with “the branch of computer science that concerns the automation of intelligent behavior. In fact, to understand what artificial intelligence is and how it can be applied in everyday life, we can think of the set of abilities of a machine to show human capabilities such as reasoning, learning, planning and creativity.
Definition of artificial intelligence
Just because of its infinite applications, that of artificial intelligence is not only a great technological challenge but also a historically and scientifically rich topic. And that refers, of course, to the intimate aspiration of man to create a machine in which his own capabilities are fully reflected. Although we are dealing with a complex technology, the basic idea of AI is very simple: to develop machines with autonomous learning and adaptive capabilities inspired by human learning models.
There are even those who point out that there is something anomalous in the expression artificial intelligence, so much so as to consider it an oxymoron, i.e. the union of two concepts with opposite meanings that, together, produce a paradox: to attribute to the artificial a prerogative typical of human nature, i.e. intelligence? Absurd!
This is not a linguistic problem that John McCarthy must have had when, in 1956, he coined the term “artificial intelligence” during a two-month seminar at Dartmouth College in Hanover (New Hampshire, USA), to which he invited ten researchers interested in the theory of automata, neural networks, the study of intelligence, but with interests ranging from the development of automatic reasoning systems to games such as checkers. Studies certainly confined – at least between the ’60s and ’70s – to the academic field, but they have played a key role in the affirmation of AI in different application areas.
After all, in modern times, artificial intelligence allows systems to understand their environment, relate to what they perceive and solve problems, so as to act towards a specific goal. The computer then receives the data – either already prepared or collected by sensors, such as a video camera – processes it and responds. In absolute terms, AI systems are capable of adapting their behavior by analyzing the effects of previous actions and working autonomously.
Weak AI and Strong AI
There are, moreover, two strands of theories on machine AI: strong AI and weak AI. We speak of weak AI when a computer, or let’s call it a robot, will never be able to reach human intellectual capabilities, but only simulate some human cognitive processes without being able to reproduce them in their full complexity. And this is the present. I fautori della teoria dell’intelligenza artificiale forte, si spingono invece oltre e ipotizzano che un giorno – magari tra solo venti o trent’anni – le “macchine” avranno un’intelligenza propria, autonoma e indipendente, pari o superiore a quella umana.
Definizioni specifiche possono poi essere date focalizzandosi o sui processi interni di ragionamento o sul comportamento esterno del sistema intelligente, ponendo come misura di efficacia o la somiglianza con il comportamento umano o con un comportamento ideale, detto razionale. Avremo così quattro principi fondamentali che possono guidarci nella comprensione di uno dei paradigmi centrali del nostro processo di ricerca e sviluppo:
- Agire umanamente: il risultato dell’operazione compiuta dal sistema intelligente non è distinguibile da quella svolta da un umano.
- Pensare umanamente: il processo che porta il sistema intelligente a risolvere un problema ricalca quello umano. Questo approccio è associato alle scienze cognitive.
- Pensare razionalmente: il processo che porta il sistema intelligente a risolvere un problema è un procedimento formale che si rifà alla logica.
- Agire razionalmente: il processo che porta il sistema intelligente a risolvere il problema è quello che gli permette di ottenere il miglior risultato atteso date le informazioni a disposizione.
In altri termini, l’intelligenza artificiale è un campo di ricerca che studia la programmazione e progettazione di sistemi mirati a dotare le macchine di una o più caratteristiche considerate tipicamente umane. Proprietà, queste, che possono spaziare dall’apprendimento alla percezione visiva o spazio-temporale. In questo scenario, capirete quanto quella dell’AI sia una disciplina fortemente dibattuta tra scienziati e filosofi poiché implica aspetti etici oltre che teorici e pratici. In 2014, Stephen Hawking warned about the dangers inherent in artificial intelligence, considering it a threat to the survival of humanity. In August of the same year, tycoon Elon Musk also expressed himself on the subject on Twitter: “We must be very careful with artificial intelligence. Potentially more dangerous than nuclear power”.
The applications of artificial intelligence
As we have already said, artificial intelligence is based on algorithms, computational techniques and solutions able to replicate human behavior. It is therefore physiological that it is now widely used in a wide variety of fields and applications, both industrial and domestic. Let’s think about medicine and robotics, but also the stock market, scientific research, law and even toys.
Not only that, within our walls there could be home automation systems capable of regulating temperature, humidity or lighting according to our needs, often using our voice as input for certain devices, which facilitate the management of our homes and, more generally, our standard of living. In the name of incisive improvement and a revolution constantly in the making.
This is possible by going to imply two basic concepts of AI itself: neural networks and fuzzy logic. The concept of a neural network is itself very simple. It is, in practice, a mathematical model – the basis of a computer system – that tries to simulate the biological neural networks of our brain where each neuron is connected on average with tens of thousands of other neurons through synapses that, in addition to allowing us to reason, also allows us to manage every function and nerve of the body. This is where fuzzy logic, also known as fuzzy logic or fuzzy logic, “hooks up”.
Everyone, or almost everyone, knows that a computer works by exploiting Boolean logic, that is, it works on two values, zero and one (binary logic). A statement, therefore, can only be true or false with no middle ground. Fuzzy logic is used in the study of artificial intelligence to introduce an intermediate truth value, i.e. a variable can take on a value of, for example, 0.2 or 0.6. An evolution of Boolean logic that, in practice, allows a given statement to be true, false or partly true or partly false.
In practical terms, now that we know what artificial intelligence is, we can move on to the more practical aspect, starting with those neural networks used in banking to prevent fraud related to the unauthorized use of credit cards. As well as in medicine to support diagnoses and better interpret medical images. Not only that, AI is widely used for the creation of online automated assistants by telephone and telecommunication companies, in order to reduce the costs of hiring and training staff.
Also in the field of transportation, the use of artificial intelligence is increasing rapidly, with fuzzy logic employed in the creation of speed changes for self-driving cars developed by giants Google and Tesla. It is also employed in the field of video surveillance, while complex neural networks are used in text generation, or rather, in the transformation of a generally textual input into an output also expressed in characters.
The Ethical Challenges of Artificial Intelligence
When we clarified what artificial intelligence is, we simultaneously hinted at its inevitable ethical challenges and the consequent cultural change that its wide use might entail. If it is undeniable that the applications are many and all decidedly innovative, there is to consider the timeless dispute between man and machine, with all the philosophical and theological implications of the case.
In fact, there are still many ethical and legal issues related to AI. Just as many are the justified doubts about what could be the impact of artificial intelligence on our society and in particular on the world of work. Among the most critical questions, one wonders if this kind of technology could represent a threat or an opportunity, if machines will gradually replace humans, and finally if the artificial systems employed will prove to be more skilled and more intelligent than the same humans who created them.
In any case, we will have to understand in the future which professional roles will be more involved in what seems to be a real technological revolution, as well as how the main job automation solutions will affect the quality of work, of the product, and the current social security system. All this while looking at a changing and bright horizon, in which innovation should represent a necessity and not a frightening threat.