What is Artificial Intelligence (AI): a complete guide


5 January 2024 - 17:00

twitter whatsapp

Everyone talks about artificial intelligence, but few know what it really is. In this article, we’ll analyze what AI is, as well as its uses and capabilities.

What is Artificial Intelligence (AI): a complete guide

In technical terms, Artificial Intelligence (AI) is the simulation of human intelligence by a digital or mechanical device. AI received renewed interest last year as new unbelievable software like ChatGPT were released to the public. However, not all of the companies that claim to use AI do. Instead, they use it as a marketing tool to attract investors and customers.

As a matter of fact, the concept of artificial intelligence today has become very broad. It’s an umbrella term that brings together different technologies, generating confusion on what artificial intelligence really is.

What is generally referred to as artificial intelligence is only a component of AI. This is the case, for example, of machine learning. This is far from sci-fi-like, ultra-intelligent Terminator machines capable of destroying humanity.

In this article, we’ll go into more detail to give a proper definition of artificial intelligence, as well as its main areas of application today.

What is artificial intelligence

A simple definition of artificial intelligence is the ability of a computer or a robot to perform tasks usually performed by humans, while also mimicking the problem-solving and decision-making skills of the human brain.

Some of the human features that artificial intelligence can imitate include speech recognition, machine learning, decision-making, visual perception, and translation.

Artificial intelligence: the origin of the term

The term artificial intelligence (Artificial Intelligence) was coined by John McCarthy (1927-2011), an American computer scientist and cognitive scientist among the founders of this discipline. In a 2005 article entitled What is artificial intelligence, McCarthy tried to answer the fateful question that many kept asking him. Here McCarthy writes:

“Artificial intelligence is the science and engineering of making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI must not limit itself to biologically observable methods.”

When asked “Does artificial intelligence simulate human intelligence?” the scientist replies:

“Sometimes, but not always and not even usually. On the one hand, we can learn something about how machines solve problems by observing people or simply by observing our methods. On the other hand, most work in Artificial Intelligence involves studying problems the world presents to intelligence rather than studying people or animals. AI scholars are free to use methods that are not observed in people or that involve much more computation than humans can do".

Machine learning and deep learning, what they are and why they are important for artificial intelligence

It is not possible to proceed in the discussion of artificial intelligence without first delving into two fundamental concepts, namely that of machine learning and deep learning.

In summary, it is possible to state that artificial intelligence is the ability of machines to carry out actions typical of human intelligence. But how do machines learn what to do? machine learning, which can also be translated as machine learning, is what teaches machines to do what men do, just as naturally.

The learning of these machines takes place through complex algorithms and can be divided into three main sub-categories: supervised, unsupervised and reinforced learning. In the first category, the machine is supplied with both input (i.e. data) and output, i.e. some examples that correspond to the expected result. In this way, the machine through data processing must understand the connection between input and output and obtain a general rule to be applied later. The second category, that of unsupervised learning, provides for the provision of only inputs and no example outputs. The third category is that of machine learning with reinforcement learning (or reinforced).

This is a machine learning method that is based on defining a "situation" and a "response" in a dynamic environment. In this case, the learning algorithm tries to learn to make decisions based on a reward or punishment (i.e. reinforcement) it receives based on the actions it performs.

In other words, the learning system must learn to select the right action according to the situation in which it finds itself, to maximize the reward it receives, and to decrease the punishments.

For example, you can think of an artificial intelligence that has to overcome an obstacle. With each move, she receives a reward or penalty based on whether or not she has achieved the ultimate goal. This then learns which is the sequence of actions that brings him faster to the goal. The learning process therefore involves the artificial intelligence starting with random behavior, trying different actions, and receiving positive or negative reinforcements depending on their effectiveness.

Over time, the agent learns to select the right actions to get the most reinforcement possible. deep learning is a sub-category of machine learning that focuses on using artificial neural networks with many processing layers to learn increasingly complex data representations. These models are capable of learning from large amounts of data and can be used for a variety of applications including computer vision, speech recognition, machine translation, and much more.

Types of artificial intelligence: weak and strong

Scholars distinguish two types of artificial intelligence: "weak" and "strong".

Weak artificial intelligence, also called Narrow AI or Artificial Narrow Intelligence (ANI), is trained to perform specific tasks and underlies most of the artificial intelligence that surrounds us today. For example, it enables applications such as Apple’s Siri, Amazon’s Alexa, IBM Watson, and autonomous vehicles.

The strong artificial intelligence is formed by the AGI (Artificial General Intelligence) and the ASI (Artificial Super Intelligence). AGI is a theoretical form of AI in which a machine would have human-matched intelligence and a self-aware consciousness capable of solving problems, learning, and making plans for the future. ASI would surpass the intelligence and capacity of the human brain.

Even though Strong AI is still entirely theoretical and with no practical examples currently in use, that doesn’t mean that AI researchers aren’t exploring its possibilities and developments. Meanwhile, the best examples of ASI may come from science fiction, such as HAL, the superhuman computer assistant from 2001: A Space Odyssey.

Where is artificial intelligence used

Today many sectors see the application of artificial intelligence. These include voice assistants, sales, social media, streaming services, cybersecurity, banking, and the supply chain.

Examples of artificial intelligence in the real world are the already mentioned virtual assistants of our devices such as Siri and Alexa, which through voice recognition can perform a series of activities for us. For example, they can play songs, control home appliances, make calls and answer messages, give directions, and read the weather forecast.

In the sales industry artificial intelligence is crucial. In a world where big data is increasingly central, artificial intelligence can be applied to the entire sales process online (pre-sales, sales, post-sales). In the first case, thanks to artificial intelligence, alternatives can be offered based on what the customer views. During the sales process, it is then possible to try to cross-sell or upsell, while in the after-sales process it is possible to assist the customer thanks to the chatbot implemented using artificial intelligence.

Even social media make use of artificial intelligence: after all, where data are abundant, technologies like this and machine learning are always present.

In Facebook artificial intelligence, or rather machine learning, is used for face verification and to detect facial features to tag friends. Deep learning is used to extract detail from an image using a series of deep neural networks.

On the other hand machine learning algorithms are used to design your feed according to your interests. Or again, you know when Netflix shows us the movies we might like? It’s because it uses highly accurate predictive technology based on millions of data points that have to do with our previous actions and content choices.

The more the data set grows, the more intelligent the platform becomes. In the field of cybersecurity and public safety, artificial intelligence is used in many countries around the world. In the case of information security, artificial intelligence is exploited to create predictive models capable of predicting potential threats to the IT infrastructure.

With this solution, technicians can secure the environment and avoid hacker attacks and consequent losses. One of the areas of public safety in which artificial intelligence is used is that of biometric recognition.

In some countries of the world, cameras are already used that can recognize people to sanction them if they are filmed while they are committing crimes.

As for the banking sector, many institutions are adopting systems based on artificial intelligence to provide customer assistance with chatbot, and detect anomalies and fraud; finance is turning to artificial intelligence to improve stock trading performance and increase profits.

Finally, AI can be used to make a company’s supply chain more efficient. In particular, companies can exploit this technology to find solutions that allow them to have the goods they need available more quickly and to spend less on the transport of goods. Naturally, these applications are only examples, since given the versatility of this technology it is possible to apply it in many other areas.

A new frontier: generative artificial intelligence

For some months now, we have no longer heard talk only of artificial intelligence, but of generative artificial intelligence, which represents a sort of evolution.

Generative AI relies on algorithms that are trained with large amounts of data, which allow it to recreate what it learns or even alter it. This is possible because, thanks to training, generative artificial intelligence assimilates the data that is provided to it and makes it its own.

This innovation has opened up the birth of new incredibly important tools that are revolutionizing the world of technology. In this sense, it is impossible not to mention ChatGPT, the chatbot created by OpenAI which exploits generative artificial intelligence and natural language processing technology (NLP) to answer questions posed to it in the form of prompts by users.

Another noteworthy software developed by OpenAI is Dall-E 2, which can generate images starting from a prompt entered by a user. This is possible because the software has been trained with thousands of images whose characteristics it has learned.

The algorithm on which it is based has therefore learned to recognize when it is asked for an image with certain characteristics. When this happens, the artificial intelligence receives the input, analyzes it, and provides an output, all in a matter of seconds.

Naturally, these tools need continuous training to always provide optimal results. A few days ago OpenAI announced that it is about to launch GPT-4, an even more cutting-edge generative artificial intelligence. It is really difficult to summarize all the applications that can be made of generative artificial intelligence.

In fact, with it it is possible to create from scratch videos, audios (also called deepfakes) or texts extrapolated from the images, just to name a few. This is also possible thanks to the great hype that exists towards this declination of artificial intelligence.

The main multinationals in the tech world are now battling it out in this field: Microsoft has signed an agreement with OpenAI to be able to integrate ChatGPT into its search engine, while Google has announced Bard, its chatbot.

Currently in the background is Amazon, which has not yet announced any chatbots based on generative artificial intelligence, although it is no secret that it is working on it. Every day new ways are born to use artificial intelligence, both generative and otherwise. It is therefore difficult, if not impossible, to summarize all the uses that can be made of it.

Surely in the coming years, this technology will become more and more present in our lives, undoubtedly also influencing the society in which we live and the way we work. It will be up to the regulators in charge of setting limits to this extraordinary technology which, due to its extraordinary nature, needs to be regulated.

Artificial intelligence: Europe, future and ethical issues

The ethical issues, as well as the opportunities that artificial intelligence opens up, are innumerable.

First of all, these concerns privacy, are considered the bitter enemy of artificial intelligence, since the latter often violates it. To ensure that this does not happen, increasingly constant and careful monitoring will be required by the institutions, called upon to guarantee the privacy of citizens.

The European Commission also presented a draft of its Regulation on artificial intelligence in April 2021. Numerous aspects have been addressed within the document, including that of autonomous driving, work, and, more generally, the ethical sphere. The future probably has a lot in store for artificial intelligence and all its forms, but to exploit it properly it will be essential that it is properly regulated, so as not to become potentially harmful to people.

Original article published on Money.it Italy 2023-03-24 10:01:02. Original title: Cos’è l’intelligenza artificiale

Trading online

Fai Trading Online senza rischi con un conto demo gratuito: puoi operare su Forex, Borsa, Indici, Materie prime e Criptovalute.