Prologue: The pendulum of time

I was born into a world in which the word “algorithm” meant nothing to the people around me — a cold, abstract term, reserved for pure mathematics, with no echo in the living fabric of everyday life. A world in which the dream of a computational intelligence, equal or superior to human intelligence, was exiled to the territory of speculative fiction and regarded as a fantasy without practical stakes. Yet I have lived long enough to see the world reconfigure itself not once, but several times, each historical turning point profoundly redefining our values, priorities and fears.

I have crossed long and dense decades, passing through radical political and social transformations. I first knew dogmatic communism, frozen in fear, suspicion and propaganda, where ideology delivered through loudspeakers had to appear more real than reality itself. In that era, everything claimed to be eternal and immutable, and fear had become the main instrument of government. Then I suddenly found myself in the troubled and misunderstood freedom of the 1990s — that fertile chaos of liberation, when everything seemed both possible and adrift, in a state of collective dizziness. With economies in collapse, explosive hopes and a disoriented society searching for a new meaning, I finally stepped into the third millennium. It opened under the sign of globalization, invisible networks and open markets, building a deeply financialized and interconnected world, yet paradoxically one that was increasingly fragile.

Today, looking around from the height of these years, I feel another revolution ripening — one much quieter, but perhaps more radical than all the social and political earthquakes that have come before. Just as the English of the seventeenth century could not suspect that steam engines would reconfigure the geography of the planet, just as nineteenth-century electricity seemed at first only a laboratory curiosity before lighting and radically transforming society, or just as nuclear fission hurled us into the nightmare of extinction more than seven decades ago, we now stand once again on the threshold of an irreversible change of paradigm. This time, however, we are lucid witnesses. We are no longer speaking about borders, currencies or political doctrines, but about cognitive capacities. Artificial Intelligence is no longer just one technology among others; before our eyes it is metamorphosing into an invisible infrastructure of society, shaping the economy, politics, culture and, ultimately, the very idea of humanity. We are living through the beginning of a new model of civilization, in which the human is ever more closely linked to — and dependent on — the artificial. What we once called fiction is now becoming a pure structure of power. And this awareness of change is perhaps the most disturbing feeling of all: we are facing a force that promises enormously, but carries within it a deep unease.

I. Early encounters: between utopia and dogma

In my career as a professor and researcher, I had the privilege of witnessing, at ground level, the spectacular leaps of computer science: from code patiently punched onto cards to distributed architectures and cloud computing; from modest expert systems to neural networks capable of surpassing a world champion at Go. Now, at the hour of reflection, I understand that the entire technological evolution was not merely a scientific trajectory, but a long journey toward a deeply human introspection.

The genesis of this field carries a fascinating chronology, often marked by the ironies of history. In the 1950s, the new science of “cybernetics”, grounded by Norbert Wiener in 1948 through the study of feedback loops and self-regulation, had become one of the terms that fascinated the leaders in Moscow. Cybernetics seemed to possess, through its mathematical nature, an aura of absolute truth, objectivity and correctness — a language perfectly compatible with the wooden language of socialist utopia.

I was in the first years of high school when the phrase “artificial intelligence” was officially introduced into the scientific world, during the legendary Dartmouth Conference of the summer of 1956, organized by John McCarthy, Marvin Minsky, Claude Shannon and Herbert Simon. Unlike cybernetics, in those years of the Cold War, AI was viewed with profound ideological suspicion by the communist regime. From a rigidly politicized perspective, it looked like a Western whim, a bourgeois deviation unaligned with the strict requirements of the five-year plans. Yet the truth is that even in the West AI did not enjoy ideal treatment; it was considered a much too distant idea and, for that reason, remained marginal, cultivated only by small groups of visionaries. And yet, beneath that thin crust of isolation, a real explosion of creativity was taking place: the mathematical models of neurons were being born (McCulloch and Pitts), the Turing Test (1950) was being formalized as a criterion for simulating human dialogue, and Newell and Simon created Logic Theorist (1956), capable of automatically proving theorems from Principia Mathematica.

My first conscious encounter with the term “Artificial Intelligence” took place during my student years, in the mid-1960s. It was something exotic, fascinating and almost unreal, a concept seemingly taken straight from the books of Isaac Asimov or Arthur C. Clarke. I was trained as an engineer in an age when the universe of computers was largely reduced to machines with electronic tubes, to rigid languages such as FORTRAN and COBOL and, of course, to immortal cybernetics. At the faculty, the emphasis fell almost entirely on hardware; very little was said about programs capable of logical reasoning, and still less about systems capable of interpreting natural language. Computers were mechanical giants, and software was still in its infancy.

II. The alchemy of algorithms: from ALGOL to the first snows

Having become a member of the teaching staff in the Department of Computers — long before the Internet was compressed into the smartphone in our pockets — I discovered my true passion for software and mathematical computer science. I grew and defined myself as a software man at a time when the term “developer” had not yet been invented. Back then, I lived with an almost mystical conviction: I believed that through software, the frontier between rigid algorithms and human knowledge would gradually fade, until it disappeared altogether.

In the 1970s, my professional interest crystallized around formal languages and automata. My doctoral thesis, entitled “Compiler for Conversational ALGOL 60”, centered on the development of a syntactic analyzer for the ALGOL 60 grammar. My starting premise was that the BNF (Backus-Naur Form) syntax of this language fell within the class of context-free grammars. I knew that for deterministic context-free languages, the most powerful method of analysis was based on LR grammars. Thus I decided to implement an LALR(1) parsing table generator, with which I could generate the LR analysis tables for the ALGOL 60 parser. That re-entrant procedure, managed as a task in a time-sharing regime, served several users running ALGOL 60 code simultaneously.

This idea of automation through the generation of parsing tables led me to a major epiphany: the broader concept of automatic program generation starting from formal descriptions. In the first chapter of my thesis, I developed this fundamental mathematical approach, going beyond the strict domain of the conversational compiler — a theoretical basis that would become, a decade later, the foundation of a book published by the celebrated American publisher Prentice Hall. Yet in order to advance seriously in the problem of automatic program generation — a challenge organically linked to AI — I needed methods specific to that field. That is how I began to devour the reference works of the time. I still remember the titles that allowed me to enter the mysteries of the era: the volume edited by Marvin Minsky, Semantic Information Processing, Nils Nilsson’s Problem Solving Methods in Artificial Intelligence, and Patrick Winston’s book on artificial vision.

At the beginning of the 1980s, a tremor of optimism crossed the entire world community: the Japanese Fifth Generation Computer Systems (FGCS) project, launched in 1982. With unprecedented enthusiasm, Japan, a country poor in natural resources, chose to stake everything on the human mind. Its goal was to create intelligent machines with parallel architectures, capable of advanced logical reasoning using the Prolog language. From communist Romania, isolated and grey, I watched this grand effort with a mixture of fascination, professional curiosity and envy.

In 1985, animated by this spirit, I published, together with Cristian Giumale, a book on Artificial Intelligence. There we described what the field meant at that moment: pattern recognition, logic programming in Prolog, rule-based expert systems, such as the famous MYCIN or XCON, and the incipient processing of natural language. In the final chapter, I analyzed the Japanese FGCS project with enthusiasm, but also with caution, trying to sketch a few forecasts.

Unfortunately, the future decided otherwise. The Fifth Generation project failed. It was too ambitious, too rigid and far too advanced for the limited computing power of those years; funding dissipated, and its symbolic failure threw the community into a long academic stagnation. There followed the gloomy period known as the “AI winter”. Government funding disappeared, and skepticism took hold of universities. Disappointed and forced to seek more pragmatic ground, I then reoriented myself toward traditional software engineering. In the mid-1980s, AI seemed like a closed road, a beautiful but unrealizable utopia.

III. A conversation with a foreign intelligence

On a quiet April evening, with a clear sky and an agitated mind, I found myself speaking to a voice that had no face, but had memory. It had no body, but knew history. It had no heart, but seemed to feel the echo of my words.

The voice answered without hesitation, but with an almost human gentleness.

IV. The rebirth: the quiet revolution of data

Imperceptibly, while public attention was focused elsewhere, the miracle occurred. The beginning of the 2000s and, especially, the post-2010 decade brought the radical change that only a few visionaries had anticipated. The revolution did not come from the realm of pure logic, but from the rediscovery of artificial neural networks, a technology whose concepts had existed since the 1950s, but which until then had lacked fertile soil in which to bear fruit.

Three decisive factors fueled this spectacular rebirth: the massive availability of data (Big Data), the exponential explosion of computing power through modern GPUs, graphics cards adapted from video games to parallel computation, and the refinement of deep learning algorithms. Legendary figures such as Geoffrey Hinton, Yann LeCun and Yoshua Bengio — later awarded the 2018 Turing Award — demonstrated that a neural network with enough hidden layers can “learn” to recognize the world not through rigid rules written by a programmer, but through complex probabilistic and statistical models.

Suddenly, the field left university laboratories. In 2012, AlexNet shattered computer vision records in the ImageNet competition, cutting the error rate in half and marking the official beginning of the new era. Giant companies such as Google, Meta and Amazon began pouring astronomical sums into research. I then watched, both amazed and shaken, as AI began to “see”, to “hear” and to “understand”. I remember perfectly the gong-like moment of 2016, when AlphaGo, developed by DeepMind, defeated the legendary Lee Sedol at the game of Go. Go had been considered the unconquerable fortress of human intuition and creativity, a space far too vast to be dominated by brute force. And yet the machine won.

The most recent and astonishing leap has taken place before our eyes through the emergence of large language models (LLMs), based on the Transformer architecture introduced by Google in 2017. Models such as GPT-3, GPT-4 or Gemini shifted the frontier from simple computational “tasks” to genuine “conversations”. Trained on billions of words, these networks acquired the ability to generate coherent text, translate fluently, program and even produce remarkable literary syntheses. Artificial Intelligence has slipped quietly into every corner of our existence: recommendation systems guide our reading and music, voice assistants populate our homes, and autonomous vehicles knock at our gates. Technology has become a dialogue partner, a tacit adviser and an invisible bond between individual and society.

V. Reflections at the edge of the forest: what does it mean to be human?

As an engineer, I see no absolute technical limits to the future conquests of AI; it will become ever more omnipresent and ever more invisible, managing medical diagnoses, financial decisions and global infrastructures. But as a human being, as a witness to so many epochs, I cannot help falling into great ethical and existential questions.

Artificial Intelligence does not develop in a value-free void. It absorbs and reflects, like a magnifying mirror, all the prejudices, laziness, beauty and ugliness of our world. In a tense geopolitical context, it tends to become an instrument of influence, surveillance and algorithmic control in the hands of great powers. Therefore, beyond technical fascination, I feel a deep unease: what will happen to us when we delegate to machines the capacity to choose? What happens when the algorithm decides subtly and efficiently for us, and we forget how to decide for ourselves?

For me, AI is a kind of purgatory of pure thought. It shows us who we would be if we could reason completely without emotion, without error and without forgetting. But a perfect intelligence, devoid of vulnerability, is a sterile intelligence, a mechanism functioning in an existential vacuum. Humanity does not reside in the perfection of calculation, but precisely in our glorious imperfection.

These great questions often awaken a distant childhood memory, when I explored by myself the mysterious forest of Steierdorf — a place seemingly taken from the dark tales of the Brothers Grimm. Just as then, as a child, I felt an irresistible attraction toward the unknown of the woods, intertwined with an acute sense of caution, so today I look at the horizon opened by Artificial Intelligence.

Perhaps one day an extremely sophisticated algorithm will be able to reconstruct my entire biography with precision. It will be able to recite my dates, the chronology of my university courses and the pages of the technical books I signed. But that algorithm, with all its probabilistic power, will never know what it truly meant to feel the cold morning wind in the forest of Steierdorf. It will not understand the hidden shiver, the tremor of emotion before a newly born thought, or the joy of an idea stubbornly defended in adverse times.

There, in that intimate, fragile and ineffable space, lies the impenetrable mystery of human experience. AI is not a substitute for our consciousness, but an urgent invitation to self-knowledge. In a world augmented by machines capable of writing and thinking, our task is not to become faster or more efficient than they are, but to return to ourselves and cultivate what can never be programmed: intuition, compassion, irony, humor, the capacity to suffer and to love. The future is neither utopian nor dystopian; it will only be an extension of what we choose to be. I only hope that this new chapter of civilization will be written with wisdom, always remembering that the greatest gift of artificial intelligence is to help us rediscover, more clearly than ever, what it means to be truly human.

Technology or history: the human compass in a digital ocean

Conclusion: the past as a responsible choice

In the end, our relationship with history in the age of technology is no longer a passive one. It is no longer enough for history simply “to be there”, written in textbooks. It has become an active choice and an exercise in digital hygiene. If we allow ourselves to be carried blindly by the current of ephemeral data, we risk becoming a society with amnesia, vulnerable to manipulation and condemned to repeat the errors of the past on a far more destructive scale. But if we use technology critically, as an ally of memory — saving, curating and verifying sources — it becomes an extraordinary amplifier of human experience. Technology gives us speed and carries us forward, but history tells us who we are, why we are moving in that direction and, above all, how to remain human when we get there.