More than ten years ago Anu Bradford coined the term Brussels effect to demonstrate the great influence that the European Union has in the world. Now, in her next book, Digital Empires (Oxford University Press), this researcher at Columbia University in New York addresses the struggle between China, the United States, and China to control artificial intelligence. Whoever succeeds will have the 21st century at their feet.
Can artificial intelligence (AI) be regulated?
That is a good question. Very important. Europeans understand that it is not easy. This is a technology that evolves very quickly. But just because it’s not easy doesn’t mean we shouldn’t try. Until now we have regulated things that seemed very difficult such as aviation security, pharmaceuticals and vaccines.
The AI ??seems even more difficult.
Yes, but the legislator does not have to be an expert to be able to regulate it. What you need is a lot of attention.
Do you think that technology companies will cooperate with the legislator or will they maintain the philosophy of maximum profit and minimum public interference?
I don’t have much confidence in technology. They have not shown that they care about user privacy, or that they manage content to prevent misinformation or hate speech. AI opens up a great business opportunity for them. It’s natural. They are companies. At the same time, though, I don’t think these companies are interested in seeing AI become a weapon for discrimination, fraud, or other criminal activities.
Users lose confidence in them.
It is evident, and that is why we are now hearing voices within the industry itself in favor of regulation. It remains to be seen, however, whether they will put aside their appetite for profit and agree to operate under a regulation designed to put the public interest first.
At the moment they invest a lot in the pressure groups that exist in Washington and it seems that they are having success.
I think so. They want to continue self-regulating, as they have done so far. Congress seems incapable of legislating AI, content, and privacy. People, losing confidence in technology companies, ask for regulation, but political polarization prevents reaching an agreement. And the lobbies are very persistent.
In Brussels there are also many pressure groups.
Yes, but a European politician needs much less money than an American one to get elected. American politicians, in this sense, are more vulnerable to lobbies.
Does the geostrategic pulse with China influence this lack of regulation in the US?
Definitely. The United States does not want regulation to stop innovation and, consequently, make it lose its primacy in generative AI, that is, the one that can create a great variety of data, such as images, videos, audios, texts and 3D models. ChatGPT is an example.
It is the AI ??that changes production models.
And the one that will decide, to a large extent, the strategic hegemony of whoever manages to dominate it better.
The United States is in the lead.
Yes. The lack of regulation has fueled innovation. Americans tend to view technology with great optimism. But now, with AI, it’s different. Experts warn of the enormous risks we will face if there is no regulation.
The EU leads regulation. By the end of the year you can pass the first legislation on AI.
The United States is watching this process closely. There is an interest in Washington and Brussels to align positions because they are concerned about China. They fear it will export AI-based surveillance technology, especially to the global south.
China imposes regulation from above, from the state, a very popular model, not only in autocracies, but also in countries of the South.
These countries need a path to technological development and China provides it. Their technology is quite good and, moreover, cheap. The United States and the European Union have no basis for asking these countries to give up Chinese technology. They offer nothing in return. And China has already built a digital silk road. Their economic diplomacy is much better than the European and North American ones.
Europe also conditions the commercial relationship to respect certain values.
These countries cannot afford to address issues such as user privacy. Protecting it requires an investment that is not a priority for them. The priority is security and many people in these countries are willing to sacrifice privacy for security. Surveillance technology reduces crime.
The world becomes more authoritarian. China can win the battle in the global south.
These countries think they can have the best of both worlds: control and growth. China has shown that you can innovate without freedom. It’s hard for us to admit, but it’s true. The global south, moreover, is very pragmatic. Chinese technology is cheap and helps solve problems such as security.
Are we witnessing an ideological confrontation between technodemocracies and technodictatorships?
Definitely. China and the US compete for technological supremacy. Chinese AI is very good at surveillance, while American AI is strong in generative AI. China cannot compete in this arena because of censorship. If you limit access to content, you limit the data the AI ??needs to learn on its own.
The lack of freedom has a cost. Do you think, then, that the Chinese generative AI will never be superior to the North American?
I would not say never, but it is clear that he is behind. China, however, is very pragmatic. He is aware that he must find a way to balance censorship with innovation, that is, to maintain control and, at the same time, drive technological progress. It will be interesting to see if he succeeds. It depends, to a large extent, on whether he does not lose his pulse with the US.
The Xi government is very authoritarian.
And he is very stubborn. He is not going to change his position and open up the country because he does not want to risk losing control, that is, the absolute supremacy of the Communist Party. His priority is not only economic growth but social stability.
It can be an opportunity for Europe and the United States.
They should form an alliance in AI against China. It is not sustainable that, while Europe regulates, the United States does not. The Biden administration wants to regulate according to the Europeans because it is what the Americans want and also because it is necessary to offer the world an alternative to the Chinese model.
Congress, dominated by Republicans, does not see it the same.
And it is a problem.
Why is the European model so good?
Because democracy does not prosper in a digital society without rules. Disinformation campaigns are capable of bringing democracy to the brink of the abyss. The American model is based on the fact that free thought, creativity without restraints, maximizes democracy. But it doesn’t work with AI. The European model focuses on preserving the rights of individuals against surveillance and interference by companies and governments. It is better for preserving democratic structures because it protects the autonomy of citizens, their ability to make the best decisions about their lives.
Inequality contributes to the decline of democracies. Tech companies do not correct this flaw.
On the contrary. Technology accentuates inequalities. They create fortunes and big gaps between the winners and losers of digital transformation. I think most people living in a liberal democracy are wary of the profit-maximizing business model and advocate a more equitable distribution of wealth. The European model is the most appropriate to achieve this.
Do you think that the regulation that the European Parliament will approve on AI will spread throughout the world, confirming the theory of the Brussels effect?
Not everywhere, but the influence of the EU will be undeniable.
Because?
Because the more data you have, the better you can prepare your models. If you want to use European data to train your AI systems you must comply with the EU requirements. If you then want to sell these systems you must also comply with European standards. The only way to escape these obligations would be to give up the European data, but then the performance of your models will be inferior. AI programmers need high-quality data, just like the Europeans.
But there is also a lot of data in the United States and China that is accessible without the requirements demanded by the Europeans.
Okay, but we are 400 million Europeans, we have a lot of data on the public sector and the industrial sector, high-quality data that is needed to have the best AI models. When, in addition, the EU, for example, punishes Instagram for advertising aimed at adolescents, the Americans want the same. They don’t want Instagram to exploit their youth. They want the same protections as the Europeans. European data can create models that are more robust and less discriminatory.
Europe has, then, a great advantage.
Europe can set the course for AI. It is true that it fights hard with Google and Meta, but, at the same time, it is difficult for these companies to replace European data with data from India, China or Brazil. In addition, the European is a more lucrative user. Advertisers pay more to advertise in Europe than in India.
Europe leads regulation, but not innovation.
Playing referee can be fun, but not as much fun as playing the game. And to play you need companies that develop the technology. Europe must be a center of production, not just the Silicon Valley of regulation, as someone has said.
What should you do to get it?
First, complete the single digital market. A small European start-up has a hard time growing because they have different regulatory frameworks in each country. The capital market union must also be completed. Businesses need financing and now it is much easier to get it in the United States. Europe also has laws that penalize the bankruptcy of a company, that is, they prevent it from taking risks. If you work with new technologies, surely one day you will fail. In Europe it is impossible to raise financing when you have failed once. In Silicon Valley it is the other way around. You can raise a lot more money after you’ve failed. Finally, the Americans are much better than the Europeans at attracting global talent. Europe must have a much more proactive migration strategy to try the same.
