01.10.2026.
Distinguished hosts, Mr Ljubić, you know that you are the first reason why I am here and that our good cooperation over many decades deserves that we meet and talk, and that no book and no amount of reading can replace a face-to-face conversation and exchange of views – looking each other in the eye and feeling each other’s energy.
I am particularly pleased to be able to welcome Mr Zielinski, and I hope I will not do him any harm by saying that I am proud that the EBRD has sent us a person of such experience and, I may say, with such an approach to Serbia’s development, as I have had the opportunity to meet with him. I am confident that, through our joint work – all of us sitting here, and who cooperate with those who are not here – will respond to the honour shown to us through the selection of Mr Zielinski by demonstrating that we deserve your presence here, and that you will justify not only the confidence of those who sent you to our country, but also our expectations that together we can help make Serbia even more respected and successful. Thank you for being with us and for always organising these meetings jointly with the FIC. And I was not joking: I meant it seriously when I said upon entering the room that Matteo Colangeli was truly a wonderful breath of fresh air in terms of the seriousness and cooperation we enjoyed at all levels. I told Mr Zielinski that he has a difficult task with such a predecessor, but that he is well on his way to challenging his first place in Serbia, and I sincerely hope that, with the help of his associates, he will succeed.
Allow me to remind all of you who were here – some believed me, while others smiled mockingly – that it was precisely here, at last year’s conference, that I referred to artificial intelligence as so-called artificial intelligence. Today, I find that many people are quoting this. Today, the distinguished Sir Roger Penrose, a British mathematician, physicist and philosopher and the 2020 Nobel Prize laureate, also says that artificial intelligence is not merely a poor term, but an incorrect one, because artificial intelligence does not exist. What do exist are mathematical models and algorithms. But what is inherent in a human being – consciousness, creativity and understanding – cannot be achieved by any machine. This does not mean that we should not use all the advantages of so-called artificial intelligence, which across the ocean has already been termed superintelligence. But the name is not what matters; the substance does – we must be aware that the responsibility we bear and the creativity that rests with us cannot be entrusted to any machine.
Before I move on to the formal part of my address, I would like to remind you of modern education, when pupils and students, instead of giving written answers, were required to tick boxes containing answers offered to them. I think that this was when the restriction of creativity and the reduction of the human being to a machine began: accepting an answer that had been offered, thereby diminishing one’s ability to think, to understand and, above all, to be authentic, distinctive and oneself – to give one’s own answer, rather than simply “yes” or “no”. That was already the beginning of a process indicating that we would arrive at the point where we are today, when we must consider how the trend of so-called artificial intelligence must not take over the role that belongs, and can belong, only to a human being.
It does not matter who is a believer and who is not, or who believes in God. Today is a wonderful feast day in the Christian religion – the feast of Love, Faith and Hope and their mother Sophia. Sophia is wisdom. God is one, and all of us here who believe that He is one call Him by different names, but love, faith, hope and wisdom remain what should guide us in everything we do.
If, like me, you have had the opportunity to work with people from the financial sector – banking, insurance and leasing companies – and indeed with all those who have indirect contact with the NBS, then you will understand how important it is to be and remain human, and not to place either machines or profit ahead of what may be called an honourable legacy.
Today there will be no poem or verse; this was a brief introduction. Allow me now to begin with four messages that define my view of the theme of this conference.
I can sum up these four messages in a few words: man has given machines speed and data, but he has not given them a conscience when it comes to making decisions, nor responsibility for their consequences. Your signatures and mine stand behind that responsibility. A machine can find a pattern, but a human being must find the measure.
That is precisely the essence of the topic that has brought us together today – innovation meets regulation.
That is why I do not see today’s topic as innovation versus regulation, or regulation versus innovation. I see it as a question of how we can enable, accelerate and scale up good innovation without losing the one thing the financial system cannot exist without – trust.
I have said at this conference before that it is up to us to embrace all the benefits that technology has to offer and to harness it in the service of societal progress. After all, people create technology to serve them, not the other way around. I have not changed that view.
Last year, at the Kopaonik Business Forum, I also said that the very term “artificial intelligence” contains a fundamental untruth. A machine works with algorithms and data; it can do many things faster than we can, but creativity, understanding and responsibility remain uniquely human.
I also mentioned Sir Roger Penrose, the Nobel Prize winner in physics, who said that what we call artificial intelligence today might more accurately be called artificial cleverness. Depending on the context, it can be translated as “cleverness” or “resourcefulness”.
Artificial ability, or perhaps even exceptional skill, but not intelligence. Because, as Mr Penrose and I believe, it lacks understanding. And I think recognising this distinction is important – not to diminish the significance of this new technology, but quite the opposite. It is important so that we can use innovation properly!
AI is also a trend today, and a trend in itself is neither good nor bad.
Recently, I bought a very rare book by Mr Pavić – a book for boys and girls, and for adults – in which there is a line I particularly liked. He says that a lady in a Russian museum had a nose that was no longer in fashion. Imagine what a wonderful illustration this is of the fact that even the human face has become a matter of fashion today, and that you can actually order the kind of nose you want. The book was certainly written two and a half decades ago, but our wonderful Pavić had already imagined back then that a lady in Russia had a nose that was no longer in fashion. Today, that would be an easy problem to solve.
We will not reject a good tool simply because it is new, but neither will we embrace it simply because everyone else is using it. The problem arises when being fashionable becomes more important than being useful.
For me, the measure is simple: what do citizens, society and the economy gain from it?
Paul Volcker, while the world was still coming to terms with the aftermath of the global financial crisis, identified the ATM as the most important financial innovation for ordinary people of the previous two decades.
Not because it was the most technologically sophisticated. Quite the opposite – because it solved a concrete problem for people – it saved them time and made banking services more accessible.
For me, that remains a good test of any innovation today: not how complex it is, but what it has actually changed for the user.
And I am not speaking merely in declarative terms, the NBS has demonstrated this through its actions.
When we launched instant payments in October 2018, using infrastructure we had developed ourselves, our goal was to enable money to reach its destination within seconds, at any time and from anywhere.
And these payments have been widely adopted – in the second quarter of this year alone, 34.5 mn transactions were processed through our IPS NBS system.
The number of registered mobile banking users has reached 5.4 mn, while in just one quarter, citizens and businesses made more than 75 mn payments via mobile and electronic banking.
Since May this year, Serbia has been operationally applying the SEPA Credit Transfer scheme, which has been joined by the NBS and 18 of the 19 banks operating in our country. This means faster, simpler and more affordable cross-border payments for our citizens and businesses.
And that is the answer to the first question – are we enabling users to carry out more easily tasks that previously took hours of their time? The answer is yes – time.
And that is precisely one of the key benefits I see in new technologies – saving time. As Seneca said, everything else belongs to others; only time is truly ours. We usually become aware of this too late: everything else can be eaten away by moths or taken by thieves, but the only thing that truly belongs to us is our time.
We can make up for many things, but we cannot make up for lost time.
And if we have enabled people to complete in one minute something that used to take an hour, we have given them back 59 minutes of their time and their lives.
If we have enabled a business owner to complete a procedure electronically, without having to make several trips to service counters, we have freed up time for them to focus on their business.
If technology frees our people at the NBS and commercial banks from routine tasks that a machine can perform faster, then they can focus on the things we truly need them for.
Ladies and gentlemen,
Finance is not an ordinary business. Our product is, to the greatest extent, built on trust.
Walter Bagehot, as early as 1873, saw the essence of the banking system in what he described as an unprecedented degree of trust among people.
Almost a century and a half later, technology has changed the way a bank looks, the way we make payments, and the way we assess risk, but it has not changed the pillar on which banking rests – and that pillar is trust.
Some of the “hidden causes” that can weaken that trust are new: the dependence of a large part of the system on a single technology provider; a piece of data ending up where it should not; a cyberattack that can disrupt a service for millions of people within minutes.
That is why the red line is trust. When a citizen deposits money in a bank, they expect their money to be safe. When a business sends a payment instruction, they expect it to be executed.
And that is why the financial system cannot simply say: “That is what the algorithm decided”.
Machines work with databases, models and algorithms, while people work with people. And no technological revolution will change that.
A machine can analyse millions of pieces of data much faster than a human can.
And we have to make use of all of this. It is no longer a matter of choice – it is part of the modern financial system. However, the role of the regulator is not to say how a model should be used or to say “do not use the model”, but to ensure that the model is reliable, that the institution understands its limitations, that the decision can be monitored, and that responsibility for the decision always remains where it belongs – with the institution and the people who make the decision.
Put simply, a machine will do what we have designed it to do and what we have instructed it to do.
But we are responsible for:
There is no software to which we can transfer our responsibility.
There is no algorithm that will say one day, instead of the governor, bank’s Executive Board, regulator or doctor: “I am responsible for the consequences”.
Recently, the NBS hosted a large group of institutional investors.
We started the discussion roughly as follows:
In August, Serbia's y-o-y inflation stood at 2.2%. Gross FX reserves reached a record EUR 30.8 bn at end-August, while net FX reserves were also at their record high.
Global business conditions are challenging, uncertainty is high, and in Serbia you have:
And then comes the question: “What is the secret of your success?”
I replied that there is no secret. There is simply responsibility for the work you do and excellent policy coordination within our ecosystem, and I am looking at Ognjen Popović from the Ministry of Finance.
Is this the result of a single model? No.
Is it the result of a single decision and a single measure? No.
Could some algorithm have guaranteed in 2012, 2020, 2022 or 2025 the sequence in which the pandemic, the energy crisis, geopolitical tensions, inflation shocks and disruptions in global markets would occur? No.
A machine can make a forecast based on vast amounts of data from past events. But the most difficult decisions faced by a central bank are often made precisely when something is happening for which there is no good precedent in the database.
That is when a model is needed – but it is not sufficient.
That is why, at the NBS, we have always invested both in models and in people who know when a model should be trusted and when another question – or at least one more question – needs to be asked before we make a decision for which we are responsible. And this applies to the financial sector as a whole. This is an important distinction between machine response and human wisdom.
Mr Zielinski,
During your visit to the NBS, you were impressed by the building itself, and in particular by the Ceremonial Hall, where we still make some of our most important decisions.
Your reaction stayed with me because the building, designed by Konstantin Jovanović, is one of the most significant architectural designs and achievements in 19th-century Belgrade.
Behind it were human knowledge, a sense of proportion, creativity and the desire to build something so valuable and enduring that it would outlive its creator.
Why am I saying this at a conference on innovation?
Because progress does not begin by forgetting everything that came before us. True innovation takes the best of what people already know and enables them to create something better.
A machine can generate thousands of proposals in a matter of seconds today. But human wisdom decides what is worth creating. And, even more importantly – what is worth preserving.
I do not cite these principles in order to slow down innovation. On the contrary. I have no doubt: innovation is essential.
But it must also be clear that open finance does not mean open data, but an open choice for the user.
The user is the one who decides who may access their data, which data, for what purpose and for how long. And when that consent is withdrawn, access to the data ceases.
At the same time, if data enable a new bank or a new service provider to assess a client more accurately and offer them a cheaper or better service, then technology has also created competition.
This is how open finance will demonstrate its full contribution – by offering greater choice, better service or a lower price, with the same or a higher level of data security.
An American economist warned back in the 1980s that innovators would be faster than regulators. And that is a reality the regulator must accept.
If the same financial service is provided in a new way, through a new channel or via a new intermediary, the risk has not disappeared just because we changed its name. That is why the same or comparable risk should imply a comparable level of protection – regardless of technology.
And where a solution is truly new, the regulator does not have to immediately choose between prohibition and completely free application.
In the world of instant payments, it is no longer enough for the system to be fast when everything works. We must also know what is happening when something does not work.
How quickly do we recognise an attack? Which service must we preserve? How quickly do we restore it? Can we continue to operate if the technological supplier we rely on is unavailable? You all know what it means when, in that one monopolised supplier, the key person goes on sick leave, vacation or is temporarily unavailable, which means you have to wait, and the users of your services also wait at some cash desk. And that is not theory, that is life.
Resilience is not measured by the fact that we will never have an incident. It is measured by the continuity of key services even when technology or an external supplier fails. We all know that as contingency planning.
If we as a regulator set zero risk as our goal, we would very quickly also get zero innovation. Our goal is risk that is recognised, measured and manageable. And here we come to perhaps the most difficult question for every regulator – when to react? The so-called control dilemma. While a technology is new, we can easily change it, but we still do not know all its consequences. When the consequences become clear, the technology is often already so deeply rooted that change is expensive and slow.
That is why a regulation that comes too early can bind the market to yesterday’s technology, and a regulation that comes too late can leave us with a risk that has already become systemic.
Nor is the number of pages of a regulation in itself a measure of the quality of regulation. Andy Haldane reminded us in 2012 that Basel I had 30 pages, Basel II about 350, and the documents of Basel III already span 616 pages. The Glass–Steagall Act, one of the key American responses to the Great Depression, had 37 pages. Its message was not that simple is always better, but that the complexity of regulations is not in itself a protection against the complexity of risk.
The task of good regulation is to provide answers to the questions:
That is why, for me, good regulation is not a limit to innovation, but a condition for people to trust it. And I do not want to speak of a choice between innovation and security, because I do not accept that choice. I often ask that question – why is the choice either–or?
We want both innovation and security. Both speed and reliability.
We want:
But I will not allow the price of that speed to be paid with trust, and that is my red line.
Because a central bank can change many things. We can change the instrument. We can change the procedure. We can change the technology. We can introduce a completely new payment system.
But there are things we must not experiment with. Security, system stability and trust. And when I say that – I know that almost all of us who are here today – and I say “almost” because I never say “all” – are aware of that.
Ladies and gentlemen,
Every era has its big words, because we are used to the idea that if something isn’t big, it didn’t even happen.
The Great Depression, the Great Recession, the Great Divergence, the Great Lockdown...
Today one of them is AI, tomorrow it might be something else.
Our task is not to be the last to accept the new, but neither is it to be the first just so that we can say we were first.
Our task is to be responsible.
To use every tool that can make our work better.
To use knowledge and technology to make decisions faster, yet equally high-quality.
But never to confuse the tool with the one holding it in their hand.
I know that the regulator’s job is twofold: not to close the door to what can bring progress, and not to leave unprotected what must remain secure.
Because a machine can be extremely clever, but a human must be wise.
Our obligation, as a regulator, is to preserve what no technology can take from us – responsibility for the work we do and responsibility toward the people for whom we do it.
That is why my advice is – keep a clear mind, understand asymmetries and especially the asymmetries of information at your disposal, and never forget the regularities of cyclical movement. Never allow yourself to waste time, as the most valuable resource, talking to people who are not wise enough to enrich you.
Thank you, and I wish you a successful fourth Financial Services Conference.
Governor’s office
