Early in my career, every Friday morning the corporate lawyers in our firm would gather around a huge round conference table, large enough to sit around 30 lawyers.

Sometimes there were 15 of us, sometimes 20 or 30, and occasionally the litigation lawyers joined us and there would be more than 50. The walls were lined with law books. It was a proper, traditional law firm conference room.

But what I remember most wasn’t the books. It was what happened around that table.

We brought the difficult matters we were working on. The lawyer handling the case would explain the issue and give his or her view, and then the discussion would start.

What about this? Have you considered that? What happens if the other side argues this? Could we structure it differently?

Those Friday mornings taught me an enormous amount about being a lawyer.

One thing my late mentor used to tell us has stayed with me throughout my career:

“A good lawyer isn’t the lawyer who knows the law. A good lawyer is the lawyer who knows where to find the law.”

I thought about those Friday mornings again recently after reading Bret Stephens’ provocative New York Times piece, “I’m Begging You: Never Write With A.I.”

His argument, broadly, is that by outsourcing writing to AI, we risk outsourcing part of the thinking that writing requires.

And I think there is truth in that.

If you repeatedly give a machine a blank page and ask it to think for you, there must be a risk that eventually you exercise those intellectual muscles less yourself.

But I don’t think the meaningful distinction is AI versus no AI.

I think it is cognitive outsourcing versus cognitive augmentation.

When I eventually founded my own firm, the dynamic changed. You go from being one of the lawyers around the table to increasingly being the person everyone around the table turns to.

That carries a particular responsibility.

Clients rarely come to lawyers with matters that are insignificant to them. It may be their company, their home, their investment, their family or their future. They are asking you to advise them because the answer matters.

And ultimately, your name sits behind that advice.

Over the last few years, I have found AI unexpectedly useful in creating some of the intellectual friction I learned to value around that Friday table.

It cannot replace the human interaction. It cannot replace the experience or instinct of a trusted colleague. And certainly not the professional reassurance that comes from looking another experienced lawyer in the eye and hearing, “Yes. I would approach it the same way.”

Nor would I ever want it to.

But I can put my thinking on the table and ask:

Here is what I think. What am I missing?

Challenge my assumptions. Give me the strongest argument against my position. Show me where this could fail. What other legal or regulatory issue might intersect with this? What haven’t I asked?

And sometimes, simply: No. I disagree with you. Here’s why.

That process has made me faster and considerably more organised. But, perhaps counterintuitively, I think it has also made me think more, not less.

Like many founders, my problem is rarely a shortage of ideas. It is cognitive bandwidth.

On any given day I may be moving between legal analysis, transactions, clients, commercial decisions, staff, business development and entirely new ideas. AI can take twenty thoughts competing for space in my head and help organise them into five questions that actually require judgment.

But the judgment must remain mine.

What should lawyers never outsource?

For lawyers in particular, I think the question “Should lawyers use AI?” is rapidly becoming obsolete.

A much more important question is: what should lawyers never outsource to AI?

Judgment. Responsibility. Professional scepticism. Understanding the client. Knowing when something legally possible is commercially foolish and when an unconventional solution may be exactly what the client needs.

And ultimately, being prepared to put your own name behind the advice.

As AI becomes increasingly embedded in professional life, regulation and responsible governance are essential. But responsible use should not mean technological retreat. It should mean understanding what we are asking the technology to do and what we must continue doing ourselves.

Which brings me back to that conference room.

Those hundreds of law books covering the walls never made any of us good lawyers. We still had to know where to look, what questions to ask, whether what we found actually applied, and what advice we were ultimately prepared to give our client.

AI has changed where we can look, beyond anything my mentor could possibly have imagined at the time.

It hasn’t changed who is responsible for the answer.

Soulla Dionysiou
Founder & Managing Partner, DP Law