EU AI law: “With regulation we prevent bad things”

Strict rules for the future use of AI will be applied in the European Union. Expert Lukovic says in an interview that the law is important. Although technology is actually not smart, it will greatly change our lives.

tagesschau.de: The European Union passed the world’s first law on artificial intelligence. High risk and risky applications are prohibited or regulated. In your opinion, how useful is the law?

Paul Lukovic: It’s a very good approach. Artificial intelligence (AI) is very powerful. It will affect our lives like no other technology in recent years. If we want it to change our lives for the better, it also needs to be regulated by law.

Regulation that does not regulate the technology itself, but its effects, makes a lot of sense. Because by doing so, we prevent something bad from happening without simultaneously hindering innovation and technology creation.

“Artificial intelligence can compromise security”

tagesschau.de: EU law fluctuates between applications – among other things, they are classified as risky and high risk. High-risk apps will be banned, while dangerous apps are subject to strict regulations. When do you think AI is risky and should be banned?

Lukovic: Risky and forbidden – these are two different things. AI is risky – just like any other technology – if it affects human well-being, human life and the security of some things that are also important to us in society. Especially if she does something wrong, she can jeopardize her safety.

However, AI is also capable of doing things that we basically don’t want to do. For example, some monitoring technologies such as the famous “social evaluation system”, which uses artificial intelligence systems to evaluate people’s behavior and see if they are behaving as the state wants them to. Basically, we don’t want that. So it is right that it is simply prohibited by law.

tagesschau.de: Where should the limits lie in the use of AI – for example when using it in the medical field?

Lukovic: It’s always a problem when AI does things without humans being able to step in or look at them again. This also applies in principle to the medical field. With high-risk applications, the issue is not so much whether we want to use the technology as it is the requirements that the technology must meet in order to be used safely.

AI should always be used in medicine if the use of AI increases the likelihood of a successful medical intervention and service to a human being.

“There is no real intelligence behind it.”

tagesschau.de: What exactly is artificial intelligence?

Lukovic: Artificial intelligence is nothing more than a set of computational methods and algorithms that are found to be able to do things that we previously thought were only for humans. For example, 20 years ago, artificial intelligence beat a human expert in chess for the first time. However, AI can also create complex images or pieces of music.

It is important to understand that no matter how amazing this may sound, there is no real intelligence behind it. At least not in the sense that we might understand intelligence. They are very precisely defined but often very simple mathematical methods applied to large amounts of data.

tagesschau.de: Does this mean that artificial intelligence only does what it is programmed to do?

Lukovic: It’s not that easy. In an artificial intelligence program, the so-called machine learning process, the computer is usually given a large number of examples. They make it clear what needs to be done. Then the computer is told step by step what it needs to do in order to conclude from these examples how the problem should actually be solved.

So the system is not learning in the sense that it is doing something completely independently. We’ve taught him how to infer something from the data, and he can’t do anything else.

But usually this data is so complex that we as humans can’t really say with 100% what the system will actually extract from the data. This is exactly where the big problem lies, and with it the need for regulation.

If we don’t take a good look at these datasets, these “example sets”, if we don’t build in certain security mechanisms, it can happen that we have a system that we think A has. But in fact, B does this – because we did not correctly understand the data we gave him.

The fact that artificial intelligence displaces humans is science fiction.

tagesschau.de: So don’t worry and can we continue to work with AI?

Lukovic: The fact that artificial intelligence will eventually establish a new intelligent species and replace humans definitely belongs in the realm of science fiction movies given the current state of artificial intelligence.

But it is a technology that affects more and more areas of our lives – for example the way we consume information. Or in traffic with self-driving cars. AI can control power grids and many other things. This is why regulation by the European Parliament is so important.

There is no need to be afraid, but we must use this technology wisely and with caution. We should always ask ourselves: is using technology in one place or another something that really benefits us as human beings or is it something that could put us at risk?

The conversation was led by Anja Martini, tagesschau.de

The interview has been edited and condensed for the transcript.