Economist Daron Acemoğlu: 'We need effective rules for AI'

Mr. Acemoğlu, in your book you write that new technologies such as AI are not sufficiently regulated and that most workers will be irrelevant in a dystopian future. Is that realistic?

For me, that's a nightmare scenario. But you'd be surprised if you knew how many people think it's a beautiful utopia. And in a way, that's the problem. AI will not lead to mass unemployment in the next two decades. But we have seen in the past that even if new technologies don't destroy jobs on a grand scale, they do make work less relevant and less well-paid. I'm not worried that there will be no more office workers or accountants. But I am worried they will no longer be as well paid as they are today. In the US, for example, the real incomes of workers without a university degree have been falling for decades. The use of AI will exacerbate this phenomenon if we don't set the right course.

The digital control mechanisms of the Chinese leadership immediately spring to mind. Should this be a model for democratic societies?

Absolutely not. But over the past 20 years, the Chinese Communist Party has poured billions of dollars into the development of AI and other internet tools, perfecting censorship and surveillance in the process. So there is a very China-specific approach to AI and other digital technologies. And that is a big problem. China may not be the overall AI world market leader, but it does have superiority in certain fields – such as facial recognition. We also need to see this: Technologies developed in China don't stay in China. Huawei, for example, has exported its technologies to more than 60 countries.

'The EU is setting a good example'

European and American companies also export surveillance technology. Is only China to be blamed?

No, I don't want that at all. There are also tech companies in the USA that have developed methods for collecting huge amounts of data and using it for police-like work. And without any regulatory framework. This may even be illegal, but there are police authorities in the USA that use this data anyway.

What is the status of efforts to regulate AI?

That is a complex question. The Chinese approach to this technology is questionable. Pretty much everything the Communist Party does has an element of control in it. On the other hand, China is also proving that these technologies can be controlled, albeit in a draconian way and with goals that we do not share in democratic societies. The argument that tech companies cannot be regulated therefore does not hold water. We should not emulate China. But China shows that it can be done.

Isn't there a huge difference after all? Anyone who does not follow the rules in China is put in prison.

Of course, there is a difference. But I don't think we need the threat of prison sentences. We can do it this way: We threaten companies with huge fines – in case they break the rules. Then they will probably stick to them. The EU is setting a good example in this respect. It's not as if the EU would or even could shut these companies down. These are US companies. But the EU is a very large market. This means that US companies have a major financial interest in being present in this market.

'Europe must also develop new technologies'

Are politicians already aware of the dangers posed by AI?

If we had talked about this three years ago, I would have said: US politicians have not understood the problem. That has changed significantly. Many government agencies are now very concerned about the possible consequences of the use of AI. And the EU Commission has always been way ahead of the US government in this respect anyway. This is probably also true for Germany, the EU's economically strongest member state. But the EU member states face different challenges than the USA.

In what way?

On the one hand, it plays a leading role in regulation, which is highly commendable. On the other hand, the EU must also encourage European tech companies to develop new technologies locally. I don't want to sound cynical: But one reason the EU is ahead of the US in terms of regulation is that so far it has been almost exclusively about regulating US corporations. So European politicians don't have to worry about killing the goose that lays the golden eggs. Things are different in the USA. Silicon Valley is a major contributor to the US economy, and US politicians are aware of this.

So what to do?

I am convinced that we need effective rules. And we need to abandon the naive optimism that AI will change everything for the better. That will not happen.

'Many questions are still unanswered'

What should we prepare for? What can we prepare for?

The genie is already out of the bottle. Artificial intelligence will influence many aspects of our social life and our economy. But we must not become fatalistic. We still have many opportunities to exert influence. Many questions are still unanswered: Who owns data? Who can use it and how? What are the rights of employees? How do we prevent AI from leading to mass surveillance – in the political sphere as well as in the workplace?

What can we learn from history? To what extent is technological progress responsible for prosperity?

In recent decades, both politicians and economists have liked to state: let technology run its course, everyone will benefit from it in the end. This is based on the assumption that technology increases productivity, which ultimately benefits both entrepreneurs and employees. However, this premise cannot be proven from history. Of course, there have been cases where technological progress has done a lot of good. Without the Industrial Revolution, we would not be nearly as well off today. We would not be as wealthy or in such good health. But some examples were less successful. There is no automatism that technological progress leads to greater prosperity.

Please give a positive example.

From the 1950s and 60s onwards, new production technologies with advanced machinery led to a huge change for a few decades. Productivity increased, but real wages across all demographic groups also rose by an average of 2.5 percentage points per year in the USA. In Germany, the increase was even steeper. And in both countries, social inequality did not increase but decreased. So there is no evidence for this period that technological progress has led to more unemployment and less prosperity.

'If we use AI correctly, we can do a lot of good'

And a negative example?

Think of the machine that revolutionized cotton production at the end of the 18th century. The cotton gin turned the US southern states, which had previously been completely economically isolated, into one of the most dynamic regions in the world at the time. Huge fortunes were accumulated. However, the workers in the cotton industry, the slaves, did not benefit at all. In fact, they were worse off. So in economic terms, this technological progress was a complete success. But for most workers, the first 100 years of the Industrial Revolution from around 1750 to 1850 were a terrible time. Real wages stagnated, working conditions deteriorated and working hours increased.

Which of these consequences will follow an increased use of AI?

We don't know yet, but we are at a crossroads. If we use AI, especially generative AI, in the right way, we can do a lot of good. Think about the shortage of skilled workers. There is a shortage of electricians everywhere, whose work will be much more complex in the future than it is today anyway. Generative AI could help solve this problem. Electricians would no longer have to rely solely on their own expertise, which is of course limited. Instead, they could benefit from the accumulated knowledge of thousands of electricians. And in real-time: try this method, take this part. Like this, an electrician could solve a problem very quickly.

Sounds good.

Basically, the use of generative AI goes hand in hand with a reorganization of work processes in many sectors. The healthcare sector would become friendlier to patients and also less cost-intensive. And in schools, it would be easier for teachers to look after pupils with learning difficulties. It's unbelievable that such good ideas are not being implemented.

'The media are falling for tech giants like Elon Musk'

Why is that?

The answer is simple. The tech companies don't care. They want to make money with AI – in the form of digital ads, for example. And we must not forget the ideological component. The corporations aim to generate autonomous machine intelligence through the use of computers and AI. It's not about the workers. This is a vicious ideology that is widespread in Silicon Valley.

If the tech companies don't care, it would be a task for politicians.

This brings us back to the question of regulation. Politicians cannot develop new technologies. But it can establish rules. And in cooperation with civil society, it can try to define values and standards so that technology benefits people and not just corporations.

What part does the media play in this?

The media should play an important part and raising awareness. But they don't do that. Instead, they fall for tech giants like Elon Musk and even glorify them. Critical voices, on the other hand, barely manage to make themselves heard.

'Germany has made a big mistake'

It is said that the new technologies could help with human challenges such as climate change. Shouldn't we be more open to these technologies for this very reason?

I am firmly convinced that these technologies are an important part of getting climate change under control. No one will disagree that the expansion of renewables is important. Incidentally, I also think that Germany made a big mistake when it phased out nuclear power. As a result, clean energy is being lost for the time it takes to phase out coal, oil and gas. But that doesn't mean that technology is a miracle cure. We still need rules to guide the use of technology. Again, this does not mean governments should micromanage and interfere in business decisions.

Given the rapid pace of climate change, do we still have time to do without the micromanagement of the state?

There is indeed no time to lose. But I don't believe that government micromanagement is the solution. Government bureaucrats are not experts in the field of innovation. The state should rather set clear rules much more strongly than it has done so far and push renewables, for example.

Daron Acemoğlu is Professor of Economics at the elite US university MIT. In his new book "Power and Progress: Our 1000-Year Struggle for Technology and Wealth," he and Simon Johnson describe humanity's struggle for control of technology and the distribution of wealth.