Artificial intelligence is already in your daily life
Artificial Intelligence will spread more and more in our daily lives. But, in reality, it is not something that is simply about to arrive: a great deal of technology that we use every day is already capable of processing information, recognizing situations, making predictions and making decisions without us intervening directly.
Not all of this is necessarily Artificial Intelligence. There are automatic systems, traditional algorithms and machine learning systems that are very different from one another. They do, however, have one thing in common: they have progressively transferred to machines part of the decisions that once directly required our attention.
Some concrete examples
- The systems used in hospitals can help doctors and specialists in the analysis of diagnostic images, in the management of clinical information and in the interpretation of large amounts of data. Today, language models too can be used as support tools, but they do not replace the doctor's assessment. And this is a principle that applies well beyond medicine.
- Commercial aircraft have been equipped for decades with extremely sophisticated automatic systems. Pilots can entrust many flight operations to the autopilot, while computers and sensors continuously process information relating to the aircraft and the surrounding environment. Automation has not eliminated the pilot: it has allowed the pilot to focus on decisions that require greater attention.
- Your smartphone continuously decides how to manage the network connection, which cell to use and, when you are abroad, which mobile network to use while roaming. Many of these decisions are made automatically based on information coming from the network and the device.
- A modern air conditioner does not merely mechanically maintain the temperature set on the remote control. It can adapt its operation to the detected temperature, humidity, outside temperature and changes in the environment, trying to achieve the desired result while consuming less energy.
In a normal home we find dozens of microprocessors and electronic systems. Televisions, phones, ovens, thermostats, water heaters, cars, elevators and many other objects are now capable of processing information and reacting autonomously to what happens around them.
Some of these systems are simply automatic. Others use machine learning techniques. Still others today use much more sophisticated forms of Artificial Intelligence.
What we call “AI” today and that we are used to querying directly through a question, a prompt or a conversation is therefore only a part of a much longer technological story.
Artificial Intelligence does not come from nothing
Artificial Intelligence is not magic. Nor is it a form of intelligence that suddenly appeared out of nowhere.
AI systems use enormous amounts of data and are developed through processes of training, evaluation and correction. A fundamental part of this heritage consists of what human beings have produced: texts, images, code, documents, conversations, examples, evaluations and many other forms of information.
It is therefore possible to say something very simple: AI also arises from the extraordinary amount of knowledge and activity that human beings have produced and continue to produce.
This, however, does not mean that an AI system reads the Internet as a person would, nor that what is published online automatically becomes true because many people have read or shared it.
The web has always been an environment in which valuable information, errors, opinions, advertising, propaganda and content of very different quality coexist. Search engines have over time developed increasingly sophisticated systems to try to identify the most useful information. Generative AI systems tackle the problem with different techniques, but they inherit the same fundamental difficulty: the quantity of available information does not guarantee the quality of the information.
Knowing does not mean understanding
This is perhaps one of the most interesting aspects of Artificial Intelligence.
An AI system can know an enormous amount of information and can produce a surprisingly articulate response. It can compare documents, summarize a text, translate a language, write code, analyze data and even hold a conversation that, at first glance, seems reasoned.
But producing a convincing answer does not necessarily mean understanding that answer in the same way a person understands it.
It is an important distinction. An AI system processes information, identifies relationships and produces results based on its own operation and the context provided to it. A person, by contrast, can also add experience, responsibility, values, intuitions, emotions and moral considerations to their decision.
Of course, human beings can make mistakes too. We can be superficial, conditioned by our prejudices or simply poorly informed. We should not, therefore, set a supposed human perfection against a supposed artificial imperfection.
The question is a different one: who makes the final decision?
A small experiment on ethics
Some time ago, for fun, I started a small discussion with an LLM. My observation began from a question that I consider anything but trivial: using content published by others to build answers and services without always clearly acknowledging the origin of that information raises issues of attribution, copyright and, more generally, of relationship with those who produced that content.
The response I received was very reasonable from the point of view of the dissemination of knowledge: using and reworking published ideas can help make them more accessible and democratically available.
I then asked a very simple question: why should the author be obliged to offer their work for free so that someone else can build a paid service on top of it?
The next response acknowledged that it was a problem still subject to discussion and regulation.
At that point I asked another question, deliberately provocative: do you know the Ten Commandments? And do you know the commandment “do not steal”?
Naturally, the system knew the concept perfectly. But knowing that a rule exists does not necessarily mean applying it as a person who understands its meaning and assumes responsibility for it would.
And this is where the experiment becomes interesting.
AI can know ethics without necessarily being responsible for the ethical consequences of its answers.
Responsibility, at least today, therefore remains largely on our side.
The real risk is not AI: it is ceasing to think
I therefore do not consider Artificial Intelligence a threat to defend ourselves against. On the contrary: it can be an extraordinarily useful tool and already today can improve a great many daily and professional activities.
The problem arises when an AI answer stops being a proposal to evaluate and becomes an answer to follow.
It is an enormous difference.
If I ask an AI system to help me organize a procedure, I can obtain in a few seconds a series of ideas that would have cost me hours of work. I can compare them, modify them, verify the information and choose the ones best suited to my situation.
If, on the other hand, I automatically consider the first answer I receive correct, I have simply transferred my decision-making ability to someone else.
And this is not a problem exclusive to Artificial Intelligence.
I remember when, as a boy, if I expressed an opinion of mine, there was a friend who every now and then asked me, with a certain enigmatic air:
“But did you read this thing, or is it just the fruit of your reasoning?”
At the time it seemed almost like a provocation. Today it seems to me an extraordinarily modern question.
Because it is a question that we should learn to ask Artificial Intelligence too:
- Where does this answer come from?
- What information is it based on?
- How important is it to verify what it is telling me?
- Which elements of my situation can it not know?
- And above all: what do I think about it?
From caution to opportunity
Refining our ability to discern therefore does not mean rejecting Artificial Intelligence. It means learning to use it better.
For an individual, and even more so for a small business, it is not necessary to become an expert in machine learning to start getting results.
AI can already help organize information, analyze documents, prepare an initial response to a customer, compare data, write or check code, create a draft of a procedure, identify repetitive activities that can be automated and much more.
True value, however, does not arise from delegating everything to AI. It arises from understanding what is worth delegating, what is worth automating and what must instead remain under human control.
This, in my view, is one of the skills that will become most important in the coming years.
Not knowing everything about Artificial Intelligence.
Knowing where it can be useful.
And, above all, knowing how to use it without giving up our ability to choose.
Because perhaps the real technological leap will not be when machines learn to do more and more things on their own.
It will be when we learn to decide, with greater awareness, which things it is truly worth entrusting to them.