Born as a scientific discipline in the middle of the last century, artificial intelligence is today indicated by industry analysts as a great technological challenge that can open up new scenarios for businesses. Around it flows rivers of money , the big IT companies do research in this sector and, if they don't have internal resources, they acquire startups or small specialized companies. Historically, the first IT companies to invest in artificial intelligence have been IBM and Microsoft , followed by players of the caliber of Apple, Facebook, Google and Amazon , just to name a few.
Currently artificial intelligence, often referred to simply as AI (artificial intelligence), finds applications in various fields, from retail to transport, from the medical sector to finance, from Internet searches to personal assistants such as Alexa and Siri.
However, artificial intelligence involves an intrinsic complexity, linked to the difficulty of defining what human intelligence is and what is meant by an " intelligent machine ". Furthermore, although numerous solutions based on artificial intelligence are available, there is still a gap between the results achieved at a theoretical level, the practical applications and the large-scale distribution of innovation potentially enabled by AI.
Artificial intelligence from its origins to today
The official birth of artificial intelligence dates back to 1956 , when a conference was held at Dartmouth College, New Hampshire on the development of intelligent machines. The initiative was proposed by a group of researchers, led by John McCarthy, who set out to create in a few months a machine capable of simulating learning and human intelligence . The challenge was welcomed by leading figures from academia and industry, including Marvin Minsky and Claude Shannon of Dartmouth College, Arthur Samuel of IBM, Ray Solomonoff and Oliver Selfridge of MIT. It was in this conference that McCarthy introduced the term " artificial intelligence " for the first timeAnd in fact sanctioned its birth as an autonomous discipline.
The goal of creating a machine capable of simulating every aspect of human learning has not yet been achieved. However, the research carried out in this direction has paved the way for new fields of study and results that, over time, have increasingly brought artificial intelligence closer to the business world. Among the milestones of this evolution are the LISP (1958), a specific programming language for artificial intelligence problems developed by McCarthy himself, and the ELIZA program (1965), which simulated the interaction between a patient and a psychotherapist.
The complex problem of building machines capable of replicating human intelligence has gradually evolved into a more pragmatic approach, based on the decomposition of a problem into sub-problems. Since the 1970s, several " expert systems " have been developed , ie programs capable of dealing with a specific problem by simulating the skills of an expert in that particular field. An important milestone in this development was MYCIN (1976), an expert system capable of diagnosing blood diseases.
It was in the 1980s that artificial intelligence left the academic world and entered the industrial world. An example of this historical step is R1 , a system used by Digital Equipment that allowed to configure orders for new computers: introduced in 1982, R1 is the first expert system used in the commercial field.
Since then, applications based on artificial intelligence have multiplied. The turning point is due to the evolution of computational capabilities and the development of a series of enabling technologies, including Big Data and cloud storage . In this development artificial intelligence is understood as a discipline that solves specific problems in well-defined areas. The approach followed is that of the weak AI , according to which machines can behave as if they were intelligent . A wide-ranging approach that keeps the aspiration for a great goal, but focuses on solving specific problems. This conception contrasts with that of the strong AI , according to which machines can actually be intelligent .
Available technologies
Over the past two decades, tools and technologies have been developed that promise companies a qualitative leap in their business management. Some solutions are consolidated and have reached market maturity, others are still under development and it is not possible to predict whether their potential will turn into a real impact for companies.
In his report : Artificial Intelligence Technologies , published at the beginning of the year, Forrester identifies the 13 technologies that it considers most significant for companies . Taking into account that the scenario is constantly evolving, the research company lists them starting from those that are applied only in very specific areas, moving on to the more mature ones and which can count on a consolidated ecosystem of suppliers, system integrators and customers.
Deep learning platforms: these algorithms are used to recognize objects in images, analyze sound waves to convert speech into text or process language and translate it into a format suitable for analysis.
Currently artificial intelligence, often referred to simply as AI (artificial intelligence), finds applications in various fields, from retail to transport, from the medical sector to finance, from Internet searches to personal assistants such as Alexa and Siri.
However, artificial intelligence involves an intrinsic complexity, linked to the difficulty of defining what human intelligence is and what is meant by an " intelligent machine ". Furthermore, although numerous solutions based on artificial intelligence are available, there is still a gap between the results achieved at a theoretical level, the practical applications and the large-scale distribution of innovation potentially enabled by AI.
Artificial intelligence from its origins to today
The official birth of artificial intelligence dates back to 1956 , when a conference was held at Dartmouth College, New Hampshire on the development of intelligent machines. The initiative was proposed by a group of researchers, led by John McCarthy, who set out to create in a few months a machine capable of simulating learning and human intelligence . The challenge was welcomed by leading figures from academia and industry, including Marvin Minsky and Claude Shannon of Dartmouth College, Arthur Samuel of IBM, Ray Solomonoff and Oliver Selfridge of MIT. It was in this conference that McCarthy introduced the term " artificial intelligence " for the first timeAnd in fact sanctioned its birth as an autonomous discipline.
The goal of creating a machine capable of simulating every aspect of human learning has not yet been achieved. However, the research carried out in this direction has paved the way for new fields of study and results that, over time, have increasingly brought artificial intelligence closer to the business world. Among the milestones of this evolution are the LISP (1958), a specific programming language for artificial intelligence problems developed by McCarthy himself, and the ELIZA program (1965), which simulated the interaction between a patient and a psychotherapist.
The complex problem of building machines capable of replicating human intelligence has gradually evolved into a more pragmatic approach, based on the decomposition of a problem into sub-problems. Since the 1970s, several " expert systems " have been developed , ie programs capable of dealing with a specific problem by simulating the skills of an expert in that particular field. An important milestone in this development was MYCIN (1976), an expert system capable of diagnosing blood diseases.
It was in the 1980s that artificial intelligence left the academic world and entered the industrial world. An example of this historical step is R1 , a system used by Digital Equipment that allowed to configure orders for new computers: introduced in 1982, R1 is the first expert system used in the commercial field.
Since then, applications based on artificial intelligence have multiplied. The turning point is due to the evolution of computational capabilities and the development of a series of enabling technologies, including Big Data and cloud storage . In this development artificial intelligence is understood as a discipline that solves specific problems in well-defined areas. The approach followed is that of the weak AI , according to which machines can behave as if they were intelligent . A wide-ranging approach that keeps the aspiration for a great goal, but focuses on solving specific problems. This conception contrasts with that of the strong AI , according to which machines can actually be intelligent .
Available technologies
Over the past two decades, tools and technologies have been developed that promise companies a qualitative leap in their business management. Some solutions are consolidated and have reached market maturity, others are still under development and it is not possible to predict whether their potential will turn into a real impact for companies.
In his report : Artificial Intelligence Technologies , published at the beginning of the year, Forrester identifies the 13 technologies that it considers most significant for companies . Taking into account that the scenario is constantly evolving, the research company lists them starting from those that are applied only in very specific areas, moving on to the more mature ones and which can count on a consolidated ecosystem of suppliers, system integrators and customers.
Deep learning platforms: these algorithms are used to recognize objects in images, analyze sound waves to convert speech into text or process language and translate it into a format suitable for analysis.
- Natural language generation (NLG) : this set of technologies enables a fluid interaction with human language to offer information, insights and interactions through long sentences or texts. They are also used to produce texts that can be read by a human being, typically starting from a body of answers or textual components.
- Swarm intelligence : swarm intelligence technologies (literally swarm intelligence) are decentralized systems to which different actors, both human and software , contribute , each of which offers a part of the solution to a problem. In this way a superior intelligence is built that brings together and increases the specific knowledge of the individual. These technologies use the behavior of social insects (such as bees) and are applied to model algorithms that respond to business objectives, how to manage a fleet of delivery vehicles, or give answers to specific questions, such as sports results forecasts.
- Biometric : biometric technologies enable a more natural interaction between man and machines. These technologies detect physical characteristics of the human body and include image recognition, voice, body language.
- Image and video analysis: these are tools and technologies that analyze images and videos to detect objects and / or characteristics of objects. These platforms find applications in various sectors, including retail, insurance, security, marketing.
- Semantic technology : a central problem for AI is to understand the environment and the context in which it is applied. Semantic technologies respond to this problem by offering a deep understanding of the data and creating the basis for introducing classifications, taxonomies, hierarchies, relationships, models and metadata.
- Speech recognition (speech recognition): these are tools and technologies that understand and interpret spoken language by capturing audio signals and transforming them into written text or other data formats that can be used in various applications, such as voice systems for customer service, mobile applications or physical robots.
- Hardware optimized for AI : this category includes GPUs and appliances designed specifically to perform specific AI tasks, such as machine learning and deep learning.
- Machine learning (ML) : machine learning platforms offer algorithms, APIs, development tools to design, develop and train models in applications, processes and other machines. The ML platforms are used in situations where, to solve a problem, it is necessary to recognize patterns within large data sets.
- Robotic process automation (RPA) : RPA technologies include various methods to automate human actions and make business processes more efficient
- Text analysis and natural language processing (NLP) : this category includes tools able to understand and interpret written text and entire documents. In the more advanced versions these tools can be used to understand emotions, feelings and, within certain limits, predict the user's intentions.
- Virtual agents : software that offers an interface that allows the user to interact in a natural way with a machine or a computer system. Among them are the chatbots widely used for costumer services and mobile applications.
- Decision management : these are software that allow you to automate decisions in real time by directly entering policies and rules that allow AI systems to deduce decisions and take action.
Among these is Open AI , which has set itself the ambitious goal of creating an AGI (artificial general intelligence), or a system capable of equaling human intelligence. In a certain sense it is a return to the original conception of artificial intelligence, the strong AI, but with the support of current knowledge and future developments enabled by the "intelligent systems" implemented to date.
Open AI, in turn, is part of Partnership on AI , an association that is proposed as a meeting place for all those who work in the AI, from the academic to the industrial world to the political one. Its members include Apple, Google, Facebook, IBM, Microsoft.
