Universal access to artificial intelligence tools does not coincide with a genuine democratization of knowledge
Abstract: This article critically analyzes the narrative of artificial intelligence as an intrinsically democratic technology, highlighting the gap between the mere availability of the tool and its actual capacity to produce emancipation, competence, and real opportunities. Starting from the public rhetoric that presents generative AI as accessible, intuitive, and capable of transforming anyone into a professionally competitive subject, the contribution shows how this representation overlooks the persistence of cultural, cognitive, and institutional inequalities. The dialogical interface of AI systems, although it appears simple and reassuring, in fact requires the ability to question, verify, contextualize, and exercise critical judgment, without which users risk confusing the formal plausibility of the output with its substantive reliability. Recent Italian judicial cases in Rovigo and Syracuse demonstrate, in the forensic field, how the unverified use of artificial intelligence can produce serious errors, non-existent citations, and professional liability, confirming that competence cannot be replaced by the linguistic fluency of the machine. The article also addresses the issue of public administration, where the use of AI encounters specific limits connected to data protection, document confidentiality, administrative responsibility, and regulatory compliance. It follows that AI will be able to become a truly democratic tool only if accompanied by critical training, adequate public infrastructures, clear rules, and governance capable of transforming technological access into effective awareness, preventing the promise of democratization from turning into a new form of cognitive dependency and social inequality.
Keywords: #artificialintelligence #digitaldemocracy #digitalinequality #digitalethics #governance #publicadministration #GDPR #professionalliability #generativeAI #criticalcompetence #law #OrestePetrone #ethicasocietas #ethicasocietasmagazine #scientificjournal #humanities #socialsciences #ethicasocietasupli
Oreste Patrone (1976), engineer and public-sector employee, he has worked for many years in the field of environmental protection and waste management, gaining solid experience through consultancy activities and institutional appointments. He currently deals with integrated environmental authorizations, waste treatment plants, and single authorization procedures, and is a member of the Friuli Venezia Giulia regional section of the National Register of Environmental Managers.
It is likely that artificial intelligence will remain the topic of the moment for a long time. Not so much because it is a passing trend, but because it has now become a structural technology, capable of entering very different areas of human experience: from work to writing, from research to education, and even the most ordinary aspects of daily organization.
Yet, rather than focusing on artificial intelligence itself, I would like to dwell on the narrative that accompanies it. There is one idea in particular that resonates strongly in public discourse: that of artificial intelligence understood as a democratic technology, accessible to everyone and capable of offering anyone extraordinary opportunities for economic advancement. This rhetoric is also fueled by the words of Nvidia CEO Jensen Huang, according to whom AI “will create more billionaires in the next five years than the Internet created in twenty.”
This suggestion is spreading across social media, amplified by the voices of artificial speakers and by sellers of a kind of expertise detached from talent and experience, promising to elevate anyone to the rank of specialist and indispensable resource. In a very short time and without effort, ça va sans dire.
Let us start from what is true, because it would be dishonest to deny it. Artificial intelligence simplifies certain concrete and everyday operations, such as solving a technical problem by uploading a screenshot, getting help to write an email, translating a text, organizing notes, summarizing or understanding the general meaning of a document. These are real simplifications, but there is an enormous distance between saying that a technology simplifies certain operations and saying that it automatically produces emancipation and opportunities for economic self-affirmation.
This second step is by no means obvious. Artificial intelligence, despite its dialogical and apparently friendly interface, does not function like a human interlocutor. It has its own architecture, its own internal coherence, and its own way of processing inputs and generating outputs. To obtain answers consistent with the question, it is not enough simply to speak to it: one must learn how to question it. And questioning it well — through precise, contextualized, and iterative requests — is a skill acquired over time and through cultural tools that not everyone possesses to the same degree. The simple and reassuring interface can mask this complexity, and it often does so very effectively.
In its early days, the Internet was described as a free, horizontal, democratic space: the great public square in which everyone would be able to express themselves, become informed, and participate. This was partly true, but that freedom, granted without adequate critical tools, also produced devastating effects: systematic disinformation, polarization, the erosion of the authority of sources, and a structural confusion between opinion and knowledge.
Studies on digital inequality have clearly shown that there are at least three distinct levels: material access to a technology, the competence required to use it effectively, and the ability to derive real benefit from it. With the Internet, we too often stopped at the first level, convinced that it was enough. With artificial intelligence, we risk repeating the same mistake, on a more refined scale and with consequences that are harder to see.
Two recent cases, both Italian and both brought before the courts, help us understand what is truly at stake.
The first dates back to April 2026, at the Court of Rovigo. A lawyer filed a defense brief. The judge read it and found, at the end of the document, a sentence that did not belong to forensic language: “If you want, I can continue by inserting this part into a complete pleading. Let me know.” The professional had copied the system’s output without rereading it, without realizing that the machine, in effect, was still waiting for instructions.
The second case is less grotesque, but precisely for this reason more unsettling. February 2026, Court of Syracuse. A lawyer included in a brief four precedents of the Court of Cassation, complete with quoted legal principles, apparently decisive for the argument being advanced. The judge verified them: none of those judgments existed. They were rulings that had never been delivered, constructed by the machine word by word, with the same form, the same tone, and the same apparent authority as real ones.
These are not cases of gross incompetence. They are cases in which the tool worked exactly as designed: it produced fluent, coherent, formally correct texts. The problem is that neither of the two professionals had the tools, or perhaps the awareness, to measure the distance between a plausible output and a reliable output.
This is the most disturbing paradox: not that people are excluded from artificial intelligence, since the tool is available, and this is undeniable; but that they become unaware users of it, convinced that they are gaining autonomy while in reality they are delegating judgments they are not yet able to evaluate. Professional appearance, in an already fast and overloaded society, can become far more persuasive than real competence. And at that point the problem is no longer technological: it is epistemic.
When artificial intelligence is described as a democratic tool, it is almost always forgotten that there are contexts in which access is not free at all, not out of hostility toward technology, but for serious and well-founded reasons. Public administration is one of the most relevant examples. The data processed there every day are public data, often sensitive: administrative acts, information about citizens, confidential documents. Uploading them to a generative AI platform means, in most cases, transferring them to servers owned by private entities, often outside Europe, with concrete implications in terms of data protection, legal liability, and GDPR compliance.
There are policies, circulars, and guidelines. In many public bodies there are explicit prohibitions or precise limitations. But the issue is not merely regulatory. Those who work in these contexts are in fact subject to a double pressure: on the one hand, the cultural push to use the tool; on the other, constraints that they do not always know in detail and that the dominant narrative tends to render invisible, almost as if they were unreasonable obstacles imposed by those who have failed to understand the future. That is not the case. They are necessary safeguards, and their existence says something important: using artificial intelligence responsibly requires an awareness that is not acquired by osmosis merely because the tool is available.
Artificial intelligence, therefore, does not eliminate inequalities. It can reduce some of them — and it would be wrong to ignore this — but it can also amplify others, and do so silently. Those who already possess culture, method, and critical capacity will use it as an accelerator. Those who do not possess these tools risk using it passively, obtaining the appearance of competence without competence; or, worse, violating rules they did not even know they had to look for.
AI may become a truly useful and more equitable tool, but it will not be so by nature, nor will it become so through mere diffusion. It will become so only if accompanied by serious critical training — in schools, universities, and workplaces — and by public governance equal to the task: clear rules, adequate infrastructures, and the political will not to allow the speed of innovation to overwhelm the capacity to govern it. Until these conditions exist, speaking of technological democracy is, in my view, premature.
The real question, then, is not whether everyone can access artificial intelligence. The real question is who bears the responsibility for building the conditions that allow access to become something real and conscious. And as long as that question remains unanswered, the democratic narrative surrounding AI will remain, at best, a promise. At worst, a story we have already heard.
FIVE LATEST CONTRIBUTIONS BY THE SAME AUTHOR
THE MYCELIAL ANATOMY OF SOCIAL NETWORK
THE IMMOBILE HORDE OF THE DIGITAL AGE
FIVE LATEST CONTRIBUTIONS CONTRIBUTIONS ON AI
ARTIFICIAL INTELLIGENCE AND CLINICAL RISK IN HOSPITALS: FROM AUTOMATION TO GOVERNED ACCOUNTABILITY
ETHICAL GOVERNANCE OF CRITICAL SYSTEMS AND ARTIFICIAL INTELLIGENCE: THE CASE OF BUSHEHR
ARTIFICIAL INTELLIGENCE AT THE SERVICE OF HUMANITY
THE IMPACT OF ARTIFICIAL INTELLIGENCE ON CORPORATE FINANCE
FIVE LATEST CONTRIBUTIONS
MAGNIFICA HUMANITAS: THE CIVILIZATION OF LOVE IN THE AGE OF ARTIFICIAL INTELLIGENCE
ANOTHER YOUNG LOCAL POLICE OFFICER HAS DIED BY SUICIDE
DREAMING ON THE TIBER BETWEEN CULTURE, ELEGANCE AND SOLIDARITY
Ethica Societas is a free, non-profit review published by a social cooperative non-profit organization
Copyright Ethica Societas, Human&Social Science Review © 2026 by Ethica Societas UPLI onlus.
ISSN 2785-602X. Licensed under CC BY-NC 4.0