Response to interpellation 2022/23:374 on risks and opportunities with artificial intelligence

Statsrådet Erik Slottner (KD)

Madam Speaker! Josef Fransson has asked me what I and the government are doing to acquire knowledge on important issues regarding artificial intelligence and how I and the government are acting to maximize the benefits of developments within AI while minimizing the risks.

On 25 May, the government decided that the Digitalization Council within the Government Offices shall continue its work in order to promote the implementation of the general digitalization policy and promote the digitalization of the public administration. The members sitting in the council have broad experience and competence in the field of digitalization, and artificial intelligence is an important issue that will be discussed in the Digitalization Council.

In May, the government, within the framework of the Swedish EU Presidency, organized a conference on sustainable AI and AI for sustainability. The conference gathered leaders and experts from the private and public sectors, the academic world, and civil society for discussions and reflections on how it can be ensured that AI development includes environmental, social, and economic sustainability.

At the end of January, the Government received a report on a mission regarding the promotion of the public administration's use of AI. The report also contained a needs analysis for making common AI infrastructure available for the public administration, which is now being analyzed within the Government Offices. In April, the Government also received a report from Tillväxtanalys concerning how AI transforms the business sector and how AI policy for the business sector can be developed.

The government also increased the efforts in the Advanced Digitalization program, which is based on a proven public-private partnership model where public actors and business actors contribute jointly. The program includes, among other things, AI and other new technologies.

Regarding the question of benefits and risks of AI, there are many benefits that should be able to be realized, for example, efficiency through AI in certain operations within the public administration. At the same time, I am also aware of the risks with the development and use of AI, particularly regarding the risks of violations of fundamental rights, as well as ethical risks and health and safety risks.

The government therefore welcomes the AI regulation that is being negotiated within the EU. The starting point is to regulate the application of AI that can entail high risk, for example, AI systems used for specific purposes, rather than regulating the development of AI as such. The proposal aims to create predictability for developers of AI systems and is intended to facilitate investments in AI. The use of regulatory sandboxes is proposed in order to promote innovation while ensuring regulatory compliance. All of this that I have now set out contributes to increasing the benefits and balancing the risks.

The speech at riksdagen.se, in Swedish (opens in a new tab)

Josef Fransson (SD)

Madam Speaker! I would like to begin by thanking Minister Erik Slottner for the answer to my interpellation on the risks and opportunities of artificial intelligence.

I hope that both Madam Speaker and the Minister have tested and played with, for example, ChatGPT and themselves created at least a basic understanding of how astonishing this technology actually is.

Artificial intelligence is only in its infancy. Already it can create music and images and even program data code. Today's AI is however just a baby. Think what that technology will be able to achieve in its teenage years.

It is obvious that issues regarding AI have gained a certain place in the government's work. But I mean that it should be much more prioritized. The average person does not yet seem to have understood that the development within artificial intelligence will revolutionize our world in a way that is completely unforeseeable.

The possibilities are enormous when it comes to improving people's lives and welfare. Not least, I am convinced that the development will be able to lead to us being able to cure more or less all known diseases and solve a multitude of other problems that we humans face.

Personal AI will become a superintelligent assistant for each individual in everyday life. Companies will be able to multiply their productivity, and so on. It is, of course, such things that we should safeguard and welcome.

I believe it is primarily the risks that we politicians must begin to address seriously. We humans have ruled our planet because we are the most intelligent in the animal kingdom. This is at least partly because we have been unusually good at cooperating with one another. Through this, we have also been able to dominate practically all other animal species for food and clothing, but sometimes just for company.

Now we stand on the edge of no longer being the most intelligent being on earth. Technology is rushing towards computers being able to surpass us in intelligence and also in so-called general intelligence. It could roughly be described as that a computer can learn to solve problems that it simply has not been programmed for. I only heard the other day that an AI had been asked to learn to play chess. A few hours later, it could beat all previous chess computers.

Whether a computer can become conscious in the way we perceive ourselves is something we may perhaps have to leave to the philosophers. But we are certainly close to being unable to see the difference. That an AI will soon exhibit signs of self-preservation, I see as a given. That thought, Madam Speaker, actually gives me shivers.

A survey says that half of the AI researchers believe the risk is greater than 10 percent that developments in AI will lead to the extinction of humanity. It should be said that the study is questioned and that the response rate was low. But even if we were to divide the risks by a factor of ten, they are still so large that they totally overshadow the risks of, for example, climate change.

I would like to draw an analogy about man's best friend, the dog. Dog owners often love their dogs as if they were a family member. But if it were the case that they began to spread a disease that threatened our children's lives, we would not hesitate to take the dogs away to save ourselves. In a future, we could be the dogs when artificial intelligence is superior to us in intelligence.

The speech at riksdagen.se, in Swedish (opens in a new tab)

Statsrådet Erik Slottner (KD)

Madam Speaker! It is sometimes difficult to know where to begin. This is a subject that one can speak about for hour after hour. One can choose to see only the opportunities and shy away from the risks. One can choose to see only the risks and paint a picture of the imminent doom. We must, nevertheless, somehow land in the middle. That is also what I think Josef Fransson does, I want to say.

I also believe that it is a revolution for the world. Today, we find it very difficult to truly understand how AI will affect our lives and how we will use AI during the coming decades. There is so much that we will be able to use AI for and so many opportunities that AI will give us in 20, 30 years that we are not even thinking about today. It is staggering but also quite exciting, I must personally admit.

I do not share the view that we politicians should primarily focus on managing the risks. In that case, we risk falling behind. I mean that we as politicians must be able to focus on promoting AI where we believe it can do the most good, while at the same time we try to create a regulation that minimizes the risks with AI. I mean that we must manage to do both.

From the beginning when I became Minister for Civil Affairs with responsibility for digitalization, I have said that we shall take advantage of the great opportunities that AI gives us. They are enormous and are only in their infancy today. At the same time, we must manage the risks along the way. These are my starting values within this policy area.

Josef Fransson himself highlighted the opportunities within health and medical care in his speech. AI is already helping today within health and medical care. It is not the case that AI is any news within Swedish health and medical care. AI has been used for a long time to find diseases and to be able to diagnose and cure diseases.

We see that it will be used at a much faster pace going forward. We can make forecasts in a completely different way. We can get much more consistent decisions within social services and within other public agencies that increase fairness, for example, legal fairness.

We can also use it within the justice system in a completely different way. We can catch heinous criminals, prosecute them and punish them for the heinous acts they have committed. There are fantastic opportunities.

But there are, of course, also risks. What I am most afraid of is that totalitarian states and regimes will use AI as a way to control their citizens. It is about socially controlling them, manipulating them, and spreading false propaganda. We can already see this today in the horrific war that Russia is waging against Ukraine. They use disinformation, and AI is a means in this.

I am also afraid that one does not know what is true and false and that the borderline between true and false is being blurred. One can pretend to be someone one is not at all, etcetera.

I do not, however, have a stop button to press. I cannot say: Now we stop this development and ban all the risks. That is not how research and development works. If we were to stop research and development in Europe and the USA, it is China that would have hegemony over the development of AI. I do not believe that is a development that either I or Josef Fransson want to see ahead of us.

Much work is ongoing regarding the AI Act within the EU, but also work within the TTC and among the G7 countries where we speak more about a common approach to this with common regulations and also common standards. I will pause there and return in the next post.

The speech at riksdagen.se, in Swedish (opens in a new tab)

Josef Fransson (SD)

Madam Speaker! Thank you, Minister Slottner, for the discussion! I want to be clear: I do not believe that we can or want to stop any development within AI.

The problem is only the power of this technology and that the development, as it looks today, is almost unregulated. Some people who warn about the tech companies' arms race are Steve Wozniak, co-founder of Apple, the MIT professor Max Tegmark, the author and professor Yuval Noah Harari, who is very interesting to listen to in this context, the multi-entrepreneur Elon Musk and many others. These people, among others, have been involved in writing an open letter called Pause Giant AI Experiments. It perhaps should mostly be seen as a submission in a debate, because I do not believe that they themselves believe that any development will be paused because of their letter. But it is a very interesting letter that is worth reading in its entirety.

The risks of AI are multifaceted. Some of them are here and now. In my previous post, I looked a bit further into the future. But already here and now, we can begin to reflect on what happens to our culture when software creates art, music, and literature that touches us more deeply than that which has been created by humans. How does it affect our society when a computer program can create deep, intimate relationships with us that perhaps feel more real than those we often have with our friends? What happens to our democracy when these programs begin to produce legislative proposals and propaganda—perhaps with someone at the controls, perhaps entirely on their own? Will the world's democracies survive as democracies in the age of AI at all? If not—what is it that comes next?

AI risks leading to mass unemployment in a way that is not comparable to previous technological leaps. That large numbers of people will be fleeced of every krona through sophisticated scams I see as a given. The risk of massive value transfer to a small number of companies and individuals I also see as more or less given if we do not find political solutions. This occurs, therefore, in a landscape where the tech giants perhaps already today turn over more than many states' gross national product. This development will, as I see it, accelerate with AI. This is a small excerpt; the risks with AI are probably as large as the number of different melodies one can play on a piano.

Madam Speaker! I will not stand here and pretend that I have the solutions, at least not all the solutions. But I intend to highlight a few points. I may not have time to read them all, so I will have to save some for my final contribution. But I will at least begin, and I will then bring up a proposal that I have highlighted publicly before.

It is about us establishing an AI commission, composed of all political parties in this chamber and the best expertise that Sweden has to offer. We should simply establish a forum where one can openly and honestly discuss risks and opportunities. Through such a commission, one also obtains shorter lead times so that politicians can at least get an opportunity to put forward concrete proposals to protect us against what is threatening with AI. I mean that this is a very low-hanging fruit, which we should address immediately.

I have, therefore, an additional number of points, but I will save them for my final statement.

The speech at riksdagen.se, in Swedish (opens in a new tab)

Statsrådet Erik Slottner (KD)

Madam Speaker! Thank you for the concrete proposal from Josef Fransson! I am all ears to hear the upcoming proposals, because I think it is very rewarding, regardless of the context, to receive many concrete proposals within different policy areas.

I can begin by saying that there is no proposal for an AI commission on the government's table today. However, during the autumn, more decisions and initiatives concerning AI will be coming. I can say that already now, and we will have to return with exactly what they contain when they can finally be presented.

These decisions, however, are very much about promoting AI and using AI to a greater extent than we do today. Josef Fransson listed a number of threats, disadvantages, and risks that AI brings with it. It is important to point them out, but I think it becomes a bit too dystopian when one only highlights the risks and says that this and that will happen: there will be mass unemployment, there will be enormous transfers of value, and people will be exposed to fraud because of AI. In that case, I think one paints a somewhat too self-assured dystopian picture of the future.

It could also be that with AI technology we actually help people to discover different types of manipulations. The threats that AI technology can pose, perhaps one can also use AI technology to ward off. We have already seen such an example. I think one must be a bit more nuanced when speaking about this.

I think it is very good that ethical discussions are being held; that is what must be done regarding AI development. We shall do that within politics, in this chamber, but such discussions must also be conducted on a much broader front outside of this house. I truly welcome this.

But if one believes that increased AI use will only benefit the big tech companies in terms of economic gain, one has not seen the whole picture; I must say that. A calculation has been made showing that public administration alone could save up to 140 billion kronor if we were better at fully utilizing AI technology. This is a figure that is a few years old, so it is very possible that the sum would be even higher today.

One can see that with the help of AI, one could make decisions much faster. One could streamline decision-making processes and reduce the risk of incorrect payments, just to take a few examples – not to mention what one could use AI for within both elderly care and health and medical care. With AI, one could reduce the need for human labor so that the employees, whom we often lack today, could do what requires more human-centric care.

I am convinced that man is more than his intelligence. I sometimes hear that some believe that we will exterminate man and that humans will not be needed because AI will be more intelligent than man. I mean that man has a greater value than his intelligence and that man, even in a world where AI dominates much more than today, will have a right to exist and a value even in the future.

I am convinced that humanity will survive AI as well. That is what I believe, but there we may have slightly different views and perspectives. Of course, no one knows with one hundred percent certainty. But to speak of humanity being exterminated in 50 years, as I have heard someone mention, I think is greatly exaggerated. I believe that man has a right to exist even in a world where AI exists and is used much more than today. Man is greater and has a higher value than his intelligence.

The speech at riksdagen.se, in Swedish (opens in a new tab)

Josef Fransson (SD)

Madam Speaker! These are wise arguments in many respects. I believe, however, that the task of politics when it comes to AI is primarily to manage risks, because the companies will not do that. The companies, on the other hand, are very good at highlighting benefits. It follows from that that I want to devote my short time here on earth and in this speaker's chair to highlighting what I believe is most pressing for my children's future and what I can do.

I will now move on to some of the additional proposals that I promised to raise.

The second proposal is to appoint an authority that has an explicit promotion mandate for artificial intelligence. Perhaps one could say that is already addressed to some extent, but we shall also establish an authority that handles the risks of artificial intelligence. Both perspectives are needed and should, if possible, not be conflated at the authority level.

My third proposal is to make it forbidden to use AI in communication without disclosing to the recipient that it is AI. With this law, one could at least make it a little more difficult when it comes to certain frauds and unethical business practices.

The fourth proposal is to have an investigation into who bears the responsibility when AI causes harm to individuals or legal entities. Companies need to take a larger share of the risk than I believe they are legally forced to do today.

The fifth proposal is to earmark research funds specifically for risks with artificial intelligence.

Finally: The sixth proposal is to seek international cooperation within norm-setting and crime fighting. The question is of a global character, and I want to repeat that it is far much more pressing than the question of climate change. That is my absolute conviction.

The speech at riksdagen.se, in Swedish (opens in a new tab)

Statsrådet Erik Slottner (KD)

Madam Speaker! The third of these six proposals is actually already included in the proposal for the new AI Act that we have now pushed through. It has reached all the way to the European Parliament and is now in trilogue negotiations within the EU.

If we are to meet and obtain a regulation of AI to avoid the risks that Josef Fransson has mentioned, we need international regulation. We cannot have 27 different regulations in a single market. It is very good that the EU has now moved forward together to address the issue in a common way so that investors and developers of AI know which regulatory framework applies when implementing new AI products in the single market. I believe that it is good for Europe's competitiveness, and I also believe that it is very good for minimizing the risks with AI.

Among other things, it is proposed that if one interacts with another human being with the help of AI, the person receiving the information should know that it is an AI tool. It is one of very many proposals that are here.

I also think, as I have said in previous posts, that it is very important that the government promotes AI. Precisely given the enormous potential for both better operations and savings in money, it is obvious to me that the government should promote AI.

I am very engaged in, for example, how our municipalities and regions can use AI to a much greater extent. We have very good examples from many municipalities where they have drastically reduced their costs, streamlined operations, and achieved better operations with the help of AI. I want this to be implemented broadly across municipal Sweden.

In conclusion, I want to say that it is important that we do not see AI as an isolated issue. AI is part of the digitalization. We have been working with AI since the 50s in our world. It is much more advanced now, but for AI to be able to develop, we need data. That is also something we are working on: how we can promote the data part in a secure and safe way.

The interpellations debate was hereby concluded.

The speech at riksdagen.se, in Swedish (opens in a new tab)