Flip Petillion
The Institute of Company Lawyers (IBJ-IJE)
the Dutch-speaking Association of the Brussels Bar (Balie Brussel)
the French-speaking Association of the Brussels Bar (Barreau de Bruxelles)
Third joint conference
Artificial Intelligence at the heart of the legal professions:
Cross-perspectives from company lawyers and practising lawyers
Keynote Speech by Flip Petillion, Member of Balie Brussel
With or Without the Butler — Preparedness in the Age of AI
Friday 20 March 2026 at 1.45 pm
podcast : https://open.spotify.com/episode/1LJHVrFClcPsbpleuRh6B5?si=urjNjbhRQqilYq0SIzynNw
Introduction
Ladies and Gentlemen,
In whatever capacity you are attending this seminar, it is a true pleasure to address you today.
Based on the number of participants, this seminar clearly touches on a subject that concerns us all.
Artificial intelligence has already entered daily legal life. Just this week, the Imec Digimeter study confirmed that more and more people use AI and GenAI — and many trust it. The technology is here and it is here to stay.
The adoption of GenAI in business and everyday life is among the fastest we have ever seen. Smartphones took roughly five years to reach 10% adoption. GenAI tools did so in less than three.
Recent court decisions show that AI is appearing in litigation and thought of in judicial reasoning, including at the Constitutional Court and the Council of State.
For in-house counsel, practising lawyers, judges and arbitrators, AI takes a very specific form. It appears through Large Language Model-based systems. And that is not a coincidence. Our profession is built on something very particular: language. We argue with language. We regulate with language. We decide with language.
So when machines suddenly become very good at language, our profession naturally pays attention.
Artificial intelligence raises a simple question: When machines assist human judgement, who is really doing the thinking?
The Butler
Allow me a small confession.
I have always had a quiet admiration for butlers.
Don’t get me wrong. It is not that I wanted to become one. But I secretly always wished to have one.
Wouldn't you?
Someone who knows you very well, but who remains discreet. Attentive. Perfectly organised. Someone who remembers what we forget and notices what we overlook.
In the novel The Remains of the Day by Nobel Prize winner Kazuo Ishiguro, we meet one of the most famous butlers in literature. Many of you will remember the film adaptation with Anthony Hopkins as the flawless Mr Stevens.
His memory is perfect. His discipline absolute. His loyalty unquestioned. He represents professional excellence.
And yet something troubling appears. His service becomes so perfect that he stops questioning the system he serves.
The tragedy of the perfect servant is not incompetence. It is obedience without reflection.
And lawyers should never become perfect servants. Nor should the tools they use. While lawyers serve their clients, they remain independent and bound by professional rules and ethics.
Stevens served the master of the house with impeccable dedication, even when history later judged his master's political judgement rather harshly.
Perhaps the real lesson of the story is this: disciplined professionals sometimes adapt too easily.
Now why am I telling you this?
Today we are building the perfect butler. Some law firms actively participate in this.
He does not wear gloves. He runs on servers. He remembers every precedent. He structures arguments instantly. He drafts pleadings before we finish reading the file. He is efficient. Tireless. Astonishing.
And – if we are honest – he is also becoming necessary.
I had the privilege of discussing this recently with the Master of the Rolls of England and Wales. We agreed that the volume and complexity of legal work today are becoming impossible to manage without technological assistance.
So the question is not whether we should have a butler. We already do.
The real question is: How do we ensure that judgement remains upstairs with the master, in our case, the master in law?
The Limits of the Butler
Artificial intelligence often sounds convincing, even when it is wrong.
We say that AI sometimes hallucinates. In medicine that is worrying. In law it may become the beginning of malpractice.
Sometimes the systems are even a little too helpful. A meeting is summarised without anyone asking for it. Reports about private conversations suddenly appear in inboxes.
We call this shadow AI.
Tools such as Read.ai may automatically send meeting reports to participants who never asked for them.
Very efficient — if you pay the subscription fee. And sometimes surprisingly creative if you do not.
Technology promises efficiency. It also increases responsibility.
The more ‘intelligent’ the tool becomes, the more careful the professional must be.
Artificial intelligence does not only create wrong answers. It can create the illusion of certainty.
In law, confidence often sounds like competence. Even when it is not.
Regulation or Principles?
Some observers say that the digital transformation of the past twenty years
was only the appetizer. And that Artificial intelligence may well be the main course. They call for new regulation.
But our profession is already one of the most regulated on earth.
Perhaps the real question is simpler: Should we not simply apply these existing principles to the AI environment? Professional dignity. Independence. Confidentiality. Integrity. Competence.
The remarkable thing is that they survive the encounter with technology quite well.
What we probably need more is education. Education at universities. Education in professional training. Education for magistrates.
In the Netherlands, some researchers suggest banning AI from education. They are heavily criticised. And I am glad that, in our country, universities are trying to do the opposite.
However, the AI focus of Universities remains largely conceptual. Students learn what AI is — not how to truly work with it. They teach awareness. But awareness is not the same as readiness.
Perhaps, we should focus even more on one particular skill: technological due diligence. So we know more about: Who built this system? On the basis of what data? Where does the data go? What exactly have I delegated?
Guidelines
In the legal profession there is a natural reflex to respond with new rules. But regulation is not the same as preparedness, which I will discuss in a while. And sometimes, more rules simply mean that we understand less of what we are doing.
Several professional organisations have already taken important initiatives, without creating new rules. I gladly refer to the guidelines adopted by the Institute of Company Lawyers and by our federal bar organisations, proposed by working groups.
Both sets of guidelines deserve recognition. They show that the legal profession is not ignoring the transformation brought by artificial intelligence.
The guidelines for in-house counsel focus mainly on governance and compliance. They remind professionals to respect the legal framework surrounding AI (data protection, intellectual property and competition law) and to remain accountable for any document produced with the assistance of AI.
The bar guidelines focus primarily on professional ethics. They emphasise independence, competence, verification of sources and protection of confidential information.
Neither approach is superior. They simply reflect different professional realities. But together they show something important: AI affects the entire legal ecosystem.
Preparedness
Yet when we read these guidelines carefully, something else becomes visible. They explain how to use AI responsibly. But they say less about how we should prepare ourselves for a legal system in which AI will play a central role.
Asymmetries
Artificial intelligence creates not only risks of mistakes. It may also create new asymmetries.
What happens when one party in a dispute has access to powerful AI systems and the other does not? How do we maintain equality of arms?
Dependence
What happens when legal professionals become structurally dependent on systems controlled by a handful of global technology providers?
I have spent half of my career working with global players in the IT world. And I admit I sometimes feel uncomfortable when I see the strategic positions some have been taking recently – and their proximity to political power.
Risk and Regulation
But preparedness goes further than that. Preparedness begins where prediction ends.
It assumes that we cannot foresee every risk. Uncertainty is a structural feature of technological progress.
Preparedness is not about waiting for perfect regulation. It is about organising responsibility in the absence of certainty.
This is not new. Societies have always lived with risky technologies or products.
We allow cars, alcohol, financial markets, even gambling – not because they are safe,
but because we have learned to manage their risks.
We do not eliminate risk. We organise it – with licences, supervision, liability and professional responsibility.
Artificial intelligence is no different. Only the scale and speed are new.
Preparedness therefore has two dimensions. The first is technical awareness:
understanding what these systems can do, and what they cannot do. The second is professional resilience: ensuring that lawyers, judges and arbitrators remain capable of reasoning independently of the tools they use.
Because the real risk is not that AI makes mistakes. The real risk is that we stop noticing them. Or worse: that we stop reflecting altogether.
Preparedness means maintaining something very old-fashioned:
the ability to think.
Access
But it also means something else. Preparedness is not only about tools.
It is about access. Who has access to powerful AI systems? Under what conditions?
With what safeguards?
We have never treated access to high-impact tools as neutral. Driving requires a licence. Medicine requires training. Weapons require authorisation.
So why would AI be different?
If access is left entirely uncontrolled, preparedness becomes little more than damage control.
Cultural
And finally, preparedness is not only legal. It is technical. It is organisational. It is cultural.
Legal rules rarely create preparedness. They stabilise it once it exists.
Which means that the real work starts before regulation. It starts with us.
Other Rules
Of course we must also respect other legal frameworks. Data protection and the other great creature of modern law: the GDPR. Many AI systems process sensitive information. And many are located outside Europe.
Technological dependence is therefore not only a technical issue.
It is also a strategic one. Which is why it may be wise to support and invest in local –preferably European – technological ecosystems and to think strategically about digital sovereignty.
Sense of Realism
Despite all these observations, we should remain realistic.
We do need the butler.
The volume of information in legal practice has exploded. Technology helps us deal with scale. And if used correctly, it may even improve access to justice.
So the attitude should be neither pessimistic nor naïvely optimistic.
Let us recognise the benefits. Remain attentive to the risks. And continue to reflect.
Last week I asked the First President of the Court of Appeal of Antwerp, a simple question.
“On a scale from zero to ten,
where would you place the current use of AI in the justice system?
Zero meaning: nothing exists.
Ten meaning: fully integrated.”
I suggested: “Maybe one?”.
He answered: “Perhaps. At most.”
Which tells us something important. We are still at the beginning. So, we can still shape the system. With GenAI adoption growing at extraordinary speed, the time to act is now.
The Way Forward
And that brings me to the final point.
Not technology. But ethics.
For centuries, access to justice depended on access to lawyers. Tomorrow, it may increasingly depend on access to artificial intelligence. Which means that our core values become even more important: Professional dignity. Independence. Confidentiality. Competence. Integrity.
If additional rules are needed, we should develop them together. In-house counsel. Lawyers. Judges. Across the entire legal system.
And we should remain a little sceptical about dramatic claims. When we read the technology press today, we might think that AI is already replacing half the workforce. But many organisations simply hired too many people in recent years. Some of the current layoffs may therefore reflect organisational corrections rather than technological revolutions.
The Public
And we should not forget someone essential in this debate: the public.
Citizens read our rules. They form expectations about fairness, responsibility and justice.
Every professional mistake therefore affects the credibility of the entire legal system. If we do not behave wisely, others will regulate us. Politicians will gladly do it. Luckily the European Commission has already started revisiting parts of the AI Act.
Conclusions
Let me conclude with a small comparison.
Some of you may know the famous advertisement by the Swiss watchmaker Patek Philippe.
It says:
“You never actually own a Patek Philippe.
You merely look after it for the next generation.”
Perhaps something similar applies to AI.
“We may never fully master artificial intelligence.
But we remain responsible for how we use it — and what we pass on.”
Not with fear. Not with blind enthusiasm. But with responsibility.
Because justice — like a well-run house — depends on something very simple: Technology may assist us. But judgement must remain upstairs.
And yes. We do need the butler.
So let us start the conversation. Let us reflect together. And let us define the limits responsibly.
In fact, that conversation already begins today. With this conference.
Enjoy it.
Thank you for listening.
