When Artificial Intelligence Is Judged with the Reflexes of a Bygone Century

The article published by RTBF about AILawyer presents itself as cautious. At first glance, it seeks to protect the public from the possible excesses of a legal tool powered by artificial intelligence. That intention, in itself, is respectable. Caution is a requirement. In the field of law, it is even a moral obligation.

But caution must not become a cover for an incomplete, unbalanced, or intellectually over-comfortable reading of the issue.

Because when reading this article, a certain unease emerges. Not because it reminds readers that artificial intelligence can make mistakes. No serious person disputes that. The unease comes from elsewhere. It comes from the fact that a technology in motion is being analyzed with static tools of judgment. It comes from the fact that a tool is being criticized for something it never promised to be. Above all, it comes from the fact that there seems to be a shocked discovery that a legal AI assistant… is not a lawyer.

AILawyer is a tool, not a lawyer

So that is the central angle: AILawyer is not a lawyer. Indeed. No more than a thermometer is a doctor, a GPS is an urban planner, or a search engine is a library. It must be said again calmly, but firmly: AILawyer is a tool. A tool for information, guidance, pedagogy, and support in understanding. Not a black robe. Not a law firm. Not an automated pleading. Not a substitute for human intelligence. To reproach it for not being what it never claimed to be is to build an empty criticism.

AI evolves faster than the law

What is most striking in this matter is not even the criticism itself. It is the framework in which that criticism is expressed. The article’s reasoning rests on an implicit logic: because an AI tool is not perfect, it should first be viewed through the prism of suspicion. Yet that reasoning forgets a fundamental fact, perhaps the defining fact of our time: artificial intelligence is now evolving faster than the law itself.

The full weight of that sentence must be understood. Law reforms slowly. It is debated, amended, interpreted, consolidated, sometimes over several years. AI, by contrast, evolves week by week. It learns, adjusts, corrects itself, refines its responses, transforms its interfaces, improves its safeguards, and redefines its scope of use. In other words, when one takes a snapshot of an AI tool at a given moment and judges it as if it were a fixed structure, one makes a methodological error. One confuses a living organism with a dead object.

The lack of benchmarking

That is where the article stumbles most clearly. It claims to evaluate, but it does not compare. It warns, but it does not benchmark. It criticizes, but without establishing the slightest serious frame of reference. There is no perspective against other tools of the same kind. No comparative testing with other legal platforms. No structured parallel with general AI systems already used massively by the public. No shared evaluation grid. No comparison protocol. In other words: no benchmark, and therefore no real measurement.

And without comparison, what remains? An impression. A direction. A narrative. Certainly not a comprehensive evaluation.

This lack of benchmarking is far from a detail. It weakens the entire exercise. One cannot, in the same breath, claim journalistic caution while giving up the most basic tool of serious analysis: comparison. If AILawyer were to be assessed rigorously, it should have been confronted with other existing solutions, other assistants, other approaches on the market. Criteria should have been established, use cases defined, gaps measured, coherence, relevance, updates, pedagogical quality, transparency, warnings, and data protection tested, and only then conclusions drawn. Without that, the reader is not given a study. The reader is given a one-sided warning.

The choice of experts

Even more troubling is the fact that the experts called upon are not mere outside observers. One of the voices mobilized in the article comes from a figure who is himself engaged in the development of legal artificial intelligence tools. That point is mentioned, but only in passing, almost at the margins, without truly inviting the reader to grasp its significance. This is not about discrediting a person or denying his competence. It is about a basic principle of critical reading: when a market participant comments on a neighboring or competing tool, that opinion must be heard, but it must also be contextualized.

That is a basic rule in any serious field. In economics, research, medicine, and industry, one does not treat the opinion of a fully external expert in the same way as that of a professional involved in a competitive ecosystem. Both may be useful. But they are not read on the same level. Failing to make that sufficiently clear deprives the reader of an essential key to interpretation.

Error is not the monopoly of the machine

The article then insists on possible errors. An incomplete answer. A faulty source. A broken link. An outdated reference. Fine. These are legitimate remarks. Any serious technology must accept criticism. But here too, the reasoning must be carried through to the end. Where is the perspective against human error? Where is the comparison with the approximations found every day on forums, blogs, search engines, or even in informal consultations? Where is the obvious reminder that professionals themselves sometimes differ in their interpretation of a rule, a deadline, a procedure, or a line of case law?

Once again, what is shocking is not that the limitations of a tool are pointed out. It is that this is done as though error were the monopoly of the machine. It is not.

The truth is more demanding than that. An AI can be wrong. A human can too. The difference is that a well-designed AI can be corrected at great speed, at scale, and continuously. That is precisely what the article itself acknowledges in its closing movement, almost despite itself: the site has evolved, its positioning has been clarified, the privacy policy has been modified, certain presentations have been adjusted, the platform has been cleaned up and reframed. In other words, the site is evolving. And not only is it evolving, it is evolving in the right direction.

Evolving is not a fault

It is quite remarkable that this point, which should be seen as a sign of seriousness, is presented almost as an embarrassing admission. Since when did correcting, clarifying, improving, and adapting a service become a fault? What should be preferred instead? A tool frozen in its imperfections? A platform immobilized in the name of consistency? In reality, this evolution is not suspicious at all. It is exactly what one expects from a responsible technology: that it listens to relevant criticism, refines its contours, improves its readability, and strengthens its safeguards.

The impossible legal absolute

Another simplification in the article must also be addressed: the idea that the tool is being judged against a kind of impossible legal absolute. It seems to be reproached for not containing the entirety of Belgian law, all case law, all doctrine, all preparatory works, all interpretive debates. But what serious lawyer claims to carry, alone and instantly in his mind, the full living body of the law, across all its branches, movements, nuances, and contradictions? None. A jurist works with sources, research, tools, document databases, analytical reflexes, and method. The AI assistant, too, belongs to an ecosystem of tools. It is not an oracle. It is not a judge. It is not a court.

Democratizing access to law

What it brings, and that is already immense, is something else: a first level of understanding, accessibility, a simplification of language, a reduction of the symbolic distance between the citizen and the law. And at that point one must ask the real question, the one too many analyses carefully avoid: who has an interest in keeping the law opaque?

Because behind the technical criticisms, one reality remains. Thousands of people do not dare ask a legal question. For fear of cost. For fear of jargon. For fear of being judged. For fear of bothering someone. Out of simple discouragement. Between silence and access to a first level of information, there is a space. It is in that space that tools like AILawyer find their meaning. Not against lawyers, but before them, around them, alongside them, as preparation, as pedagogy, as complement.

To refuse to see that is to understand nothing about the transformation underway.

A debate that deserves better than caricature

The serious debate is not lawyers versus machines. That opposition is lazy. The serious debate concerns how to organize intelligently the coexistence of human expertise, deontological requirements, access to law, technological speed, and responsibility. It is a fascinating debate. But one must be willing to engage in it without caricature.

The substance of the matter is therefore simple. Yes, caution is needed. Yes, clear warnings are needed. Yes, transparency, confidentiality, pedagogy, and safeguards must be constantly improved. Yes, it must be repeated that no technological tool replaces, in complex cases, the individualized analysis of a qualified professional.

But no, one cannot seriously treat a legal AI assistant as though it had appeared one morning in a courtroom wearing a lawyer’s robe. No, one cannot claim to educate the public without offering a comparison of the market. No, one cannot ignore the speed at which these tools evolve. No, one cannot remain silent about the fact that some commentators themselves belong to that technological universe. And no, one cannot turn into a weakness what is in fact the defining mark of a living project: its ability to evolve.

Conclusion

In the end, perhaps the greatest paradox of this article lies there. In trying to remind people that AI must be handled with caution, it mainly reveals the extent to which some still look at this revolution with the reflexes of another era.

Legal artificial intelligence is not a lawyer. No serious person claims that it is. It is something else. And that is precisely why it is unsettling: because it opens a door that, until now, remained closed to many.

So the issue is no longer whether one should deny its existence, nor whether one should feign surprise at its imperfection. The issue is who will have the intellectual honesty to judge it in its proper place, with method, with proportion, and with the elementary lucidity that every great transformation demands.

Law, by nature, loves stability.
Artificial intelligence, by essence, lives through evolution.
Between the two, fear must not be the choice.
Intelligence must be.