Having a general-purpose AI as an advisor is a bit like having a bodyguard who gives you good advice… but isn’t out there with you on the ground.
As a legal expert working within Culturepédia’s multidisciplinary team, I’m involved in developing specialized, sovereign AI systems designed to support our cultural ecosystem. At the same time, just like my teammates, I also use generative AI based on large language models—such as Claude and Anthropic—as part of my daily work. Unsurprisingly, I’m challenged in my work by people who use AI to obtain legal recommendations. I find myself, in a way, in the position of the hoist by his own petard.
Despite their limitations, it’s clear that these advanced computing systems are gaining ground very rapidly. How does my work as a legal advisor provide real added value in the current context? What is the difference between human expertise and the recommendations we can obtain in a matter of seconds with the help of these AI systems? In a context where organizations’ financial resources are inevitably limited, this is a perfectly legitimate question. My answer boils down to two points that illustrate how a human expert, regardless of their field of specialization, stands out from a machine:
1. Accountability
No matter how advanced AI may be, it cannot sign an opinion on your behalf or assume responsibility for its recommendations. A duly authorized attorney can. If the advice they provide turns out to be unfounded, it is their reputation and career that are at stake. Of course, not all humans are trustworthy. There are more than 30,000 lawyers registered with the Quebec Bar alone, not counting the approximately 4,000 notaries registered with the Order of the Chamber. There are undoubtedly a few black sheep among them. Still, for the vast majority, we’re talking about people with careers, families, and friends. They won’t take the risk of needlessly jeopardizing their reputation. By taking up their clients’ cause, they assume personal liability—something no technological system can do directly.
2. Understanding the context and what is the right thing to do
The job of a good legal advisor, like that of a good bodyguard, consists mainly of anticipating threats and putting in place protective measures tailored to the situation. The use of weapons is sometimes necessary to counter attackers, but as anyone who has been through it knows, the outcome of a fight is always uncertain. It is therefore essential to prepare your defenses. The energy, time, and approach required depend on the context. A hundred-page contract may protect you in many ways, but will you have the time needed to negotiate each of its terms with your counterparts? To strike the right balance, one must understand the situation as a whole—from a legal perspective as well as in terms of its economic aspects and the underlying human dynamics.
A good lawyer has a broad perspective on the law and understands its fundamentals. They make it their duty to learn continuously. Through experience, they are thus able to refine their ability to do the right thing, for the right person, at the right time. To do this, one must live and have lived—in short, be vulnerable and mortal. One must combine knowledge with intuition.
So, what do we do now?
AI, whether general-purpose or specialized, is a powerful and now indispensable tool. When used judiciously, it can be a useful aid, provided that a framework for responsible use is established. The Quebec Bar Association has done remarkable work in this regard by providing its members with a well-structured initial user guide. The key principles are clear: be transparent in your use of AI; verify your sources; ensure the confidentiality of case files and the protection of information; and document and archive your query engineering.
Provided an appropriate ethical framework is followed, expert legal advisors—whether working alone or in teams—are now able to accelerate and enhance key aspects of their work. Does this pose a threat to the legal profession? I don’t think so. Rather, it’s an opportunity to expand and democratize access to legal advice. Due to budget constraints—in the cultural sector as in many others—many organizations are forced, in the heat of the moment, to do without expert advice, and this is not for lack of wanting proper guidance.
Every organization should be able to count on a good legal advisor, and everyone should have timely access to a good personal trainer, nutritionist, physical therapist, doctor, psychologist, broker, real estate agent, accountant, engineer… (I’ll leave it to you to complete the list). AI, conceived as Augmented Intelligence, is a tool that has the potential to provide more informed advice to more people. We’re far from having won this battle, but it’s certainly within the realm of possibility.
The informed use of large language models, in their current form, is one avenue to explore. The deployment of specialized, cost-effective, and “tailored” AI is another. It is unrealistic to expect to turn our backs overnight on the use of all major technology platforms for every purpose. That said, it is crucial to develop our own solutions in parallel. This is the path Culturepédia is taking: autonomous and frugal technology, specific use cases, and a robust legal protection and governance framework, reinforced by fiduciary governance. Building our own tools and spaces of freedom is the path that leads to balance.
In an upcoming article, I will explore the practical application of the arts of self-defense to the challenges and threats posed by the ever-expanding AI, and how Culturepédia draws on time-tested principles that have stood the test of time.
This article was written without the use of AI. It reflects solely the author’s opinion.
