The race for artificial intelligence - Can Europe compete? | DW Documentary

The race for artificial intelligence - Can Europe compete? | DW Documentary

Show Video

The race for artificial intelligence has taken off. As AI takes the world by storm, transforming industries and societies, researchers – and countries – are competing to develop the most advanced technology. So far, the frontrunners are the United States and China. But Europe is determined to catch up with its own vision for AI – one that’s based on stricter rules to protect user data and privacy.

If there is no trust, there is no use. We set out to understand: Will this strategy help Europe close the gap? We’re meeting officials and AI pioneers across the continent to find out. We are on one of the busiest crossroads in Sofia, in district Lozenets.

District Lozenets is our study area. We have six lidar sensors that perform laser scanning of the crossroad. We use this data to monitor the traffic on the crossroad. What’s over there? What’s that? That’s the air quality station.

It measures air pollutants, and also noise levels and rain levels, wind speed and wind directions. See the noise levels: During the weekend, there is no noise, because there is no traffic. But now, the noise level is very high. My name is Dessisslava Petrova-Antonova.

I'm leading the city digital twin pilot project. And the aim of this project is to create a digital twin of Sofia City. Petrova-Antonova and her team are working on a digital replica of a central neighborhood in Bulgaria’s capital, Sofia. To do that, they are collecting data all across the city. Now we are very close to one of the tallest buildings in the district.

This is the Hotel Marinela. This is one of our use cases here. We are using the 3D model of this building to perform 3D simulations. Data like this is used to build a virtual model of the neighborhood. That helps the researchers to understand, for example, how the wind moves through the architecture of the city, or where air pollution is particularly bad. Then, they use AI to come up with suggestions for how to make the neighborhood more livable: From ways to save energy to how to improve walkability.

And they test these suggestions in their computer model to see how effective they are before being implemented into the real world. The main purpose of the digital twin is to guide this decision-making process in the city. In the digital twin, you can simulate ‘what if’ scenarios: What will happen if I change the street, or what will happen if I construct a new building here? The mistakes should be made in the digital twin, not in the real city. The researchers we’re meeting in Bulgaria aren’t the only ones attempting to use AI to improve life in their hometown: Cities from Santiago de Chile to Hyderabad in India are working on similar initiatives.

These “smart cities” are one of a few visible symptoms of an otherwise invisible revolution: AI technology has long become a part of our world, with AI algorithms deciding now, for instance, what we see on social media. That’s how increasingly, we’re integrating AI technology into our lives… …and increasingly, it’s allowing us to reach completely new frontiers of knowledge. In Poznan, Polish research institutes are figuring out the future of biochemistry.

And the humans we meet here have a very special helping hand. We can run much, much more experiments than 10 or 20 students can do during one day. This one-of-a-kind robot operates a fully automated laboratory, running thousands of tests daily and autonomously analyzing their results.

My name is Marek Figlerowisz and I’m the director of the Institute of Bioorganic Chemistry, Polish National Academy of Sciences. Here, AI is used to work out the most effective medical treatments for certain cancers. In each tube, we have a couple of cells representing this cancer. Yes. And then to each tube, we can add a different compound. During one day, we can find ten of them which are the best candidates for therapeutics against this disease.

And that means down the line, medics will be able to rely less on trial and error. In a hospital, it is quite difficult to make decisions about whether we should apply this type of treatment in cancer or another. Artificial intelligence can help us to make decisions. So what you're saying is that artificial intelligence can help scientists to save lives? Yes, for sure.

Yeah. With the ten years of planning and multi-million-euro price tag attached, this kind of lab won’t be replacing human workers anytime soon. But on the streets of Poznan too, AI is already weaving its way through daily life, We’re traveling across town to one of the country’s top biochemistry centers. Here, scientists drawing on European Union funds are trying to shake up life science research. My name is Luiza Handschuh and I’m a scientist, deputy director of the Institute for Bioorganic Chemistry.

It all starts with what makes us, us: DNA: So it's deoxyribonucleic acid. Researchers receive human saliva or blood samples collected in hospitals and compare DNA from cardiology and cancer patients to results from the general population. The goal is to create new resources to work out what’s driving the disease and improve diagnostics.

Here, we prepare the samples for genome sequencing. So to sequence a genome is to collect all the information, all the genetic information, which is inside our bodies, inside cells. But just one genome sequence generates huge amounts of data. And big data can generate big problems. Sometimes, it's impossible to compare so many features in one test. For a human being, it's impossible.

That’s where machine learning comes in. To make that data useful, AI can crunch the numbers. These algorithms have much more data than even the best specialists. And they can suggest that maybe this disease can be connected with something particular. But things are about to get more complicated. Because like all of us, researchers will soon be using AI under a strict new rulebook being brewed up in Brussels.

In many ways, the European Union is considered the global stronghold of privacy. The bloc already has much stricter data protection rules than competitors like China and the US We’re here to find out what’s behind the bloc’s upcoming laws for the age of AI. My name is Lucilla Sioli and I work on Artificial Intelligence at the European Commission. Officials here have spent years drafting what’s considered the world’s most comprehensive legislation on AI. The idea: regulating uses according to the risks they pose. And we’re told that those risks are real: AI has been shown to replicate biases or supercharge disinformation.

Artificial intelligence is a black box. And so the way you want to mitigate this risk is to ask the developers to document their data sets, give the information to the user of that artificial intelligence system about how that artificial intelligence system works. The question now is: Do the companies developing AI technology consider this approach a blessing – or a curse? We are designed to be a game changer. That's what the [European] Commission expects from us – to change the research system in Bulgaria.

In Sofia, the “digital twin cities” project is part of a broader big data initiative called “GATE.” It was founded in 2019, in cooperation with a university in Sweden, and co-financed with 30 million euros in EU seed funding. My name is Sylvia Ilieva and I'm the director of GATE Institute. Her center is doing research on how to apply artificial intelligence to develop real-life solutions. We defined four application areas in the beginning.

In addition to future cities, other application areas are digital health. Researchers at GATE are using AI to better understand cognitive diseases like Alzheimer’s. They are also looking into how to use technology to detect and fight disinformation online.

And the last application area is ‘smart industry’ where we will start working more intensively this year and where we’ll offer some data services for companies. All of those projects rely on vast amounts of data to become effective – data that’s often difficult to obtain in privacy-savvy Europe. It's a big challenge and it's a challenge because of the people, because of the mindset of the people. And yet, Ilieva believes that the EU's approach of establishing strict rules to emerge as the world’s leader in trustworthy AI is the right path forward.

Absolutely, Yes. It's a matter of trust between us. But not everyone is convinced. We are in a technological race and the first thing that should be moving us is are we part of the technological revolution? Otherwise, the risk for Europe is to be kicked out of history. I'm Cedric O, I used to be a minister in France in charge of digital for the past three years. And now I'm a consultant.

This summer, the former politician co-wrote an open letter with over 160 European business executives, warning that the EU’s planned AI laws could jeopardize Europe's competitiveness. We, for sure, have to be able to protect our citizens from collateral damages, but we have also to ensure our citizens will still be living in a continent that has the same access to technology, that will go on with creating jobs, that will be sovereign in its technological choices. O isn’t alone: Other industry insiders express similar concerns. And yet, when we talk to digital rights advocates, they stress that protecting users’ rights should always come first. To say that this harms competition, I think it's ludicrous.

And I think it shows a very poor understanding of what protecting people means and what democracies mean. My name is Gemma Galdon-Clavell. I'm the CEO and founder of Eticas Tech. Eticas Tech is an organization that works to protect people in technology processes. Look at what's happening with ChatGPT, for instance.

ChatGPT was launched as a test, using us all as guinea pigs, using our data, going over any kind of regulation that already exists in Europe to protect, to protect people. Now they're being hit by all these lawsuits. So I just don't understand this logic of ‘because the others are doing things wrongly, we should be able to do the same thing. At the same time, experts tell us that rules won’t be the only factor determining Europe's success. Another one is money. Although private AI investment in Europe has been growing in recent years, it still lags behind China and, even more so, the US.

And the challenge is: How do you change that? If we want to get back into the race, we need investors to believe that Europe is the right field to invest in technology. So we have to be aware and be very cautious of not overregulating – otherwise, we are threatening the investment, and so we are threatening our own future. Back in Brussels, EU officials tell us their upcoming AI rules won’t place undue demands on innovators. In fact, what we ask companies to do is to simply assess before they put the artificial intelligence system to be used for a purpose that is considered high risk. And to be frank with you, a developer that develops AI for a high-risk intended purpose should do those things already.

The bloc itself invested around 4 billion euros in artificial intelligence from 2021 – 2022. Although most of today’s cutting-edge AI applications were developed elsewhere, Brussels insists Europe is well-placed in the race for AI. Everybody knows ChatGPT.

But in reality, there are also models that have been developed in the European Union and maybe they are smaller and less known, but they are there. it may be that we have a delay of about a year or so, but we are not so late and we can definitely still catch up. In Poland, people aren’t so sure.

Institutes here in Poznan have received tens of millions of euros from the European Union’s cohesion funding pot. But… Frankly speaking, it's still not enough money for this type of research, for sure. And if something is happening now, it is difficult to find new money. And our reaction can be a little bit late. Further south, Bulgaria is keen to join the AI race on more than one front. Last year, the Bulgarian government co-founded another independent unit doing fundamental AI research in Sofia called INSAIT.

This can help Bulgaria enter into direct competition with countries across Europe, North America, and Asia in the field of AI and related research. Despite its well-educated workforce, Bulgaria is still the EU’s poorest member: Every year, an estimated 30,000 Bulgarians leave the country of less than 7 million in search of better career prospects. I also had the opportunity to go abroad, but I noticed that here there are a lot of knowledgeable and skillful people and I also have a passion for developing this region.

And I would like to show that world-class research can be done here. And it's home also, I guess? Yeah, that's also a factor. It's home for me. I'm Kostadin Garov.

And currently, I'm a PhD researcher at the INSAIT Institute, where I work on the intersection of artificial intelligence and security. Garov is doing research into an AI method called “federated learning.” It’s supposed to allow institutions to share data to train AI models without revealing sensitive information.

But when Garov and his colleagues simulated a cyber attack, they managed to recreate much of the data that was thought to be safe. This is the original image and this is the image which was reconstructed by the server. And these stolen reconstructions are realistic enough that they can be considered a privacy breach. This is the data that people thought was private, was private and safe? Yes, exactly.

In our simulation, the thief, the attacker, the server manages to steal these representations of the data, which are not perfect reconstructions, but they are revealing enough to be considered a threat – a serious threat. The leaders of the institute tell us that fighting these new threats is one of the center’s goals. And to make this happen, they need to attract the right talent. Once you get the best people, everything else follows. My name is Martin Vechev. I'm a full professor at ETH Zurich in Switzerland and lately, I'm the founder of INSAIT.

Vechev’s own research focuses on quantum computing, the next-generation technology set to revolutionize the way computers work. He too left Bulgaria early in his life to study and work at universities and tech companies in the West. It's very common for Bulgarians to go abroad. It is a substantial problem because you have some of your most talented people leaving the country. So this is a direct hit on the economy.

Vechev and his colleagues are determined to reverse this trend. And to achieve their goal, they’re also collaborating with Big Tech companies such as Google and Amazon, many of which have their headquarters outside of Europe. Big Tech companies like Google, like DeepMind and AWS have also a lot of world-class research going on. So for new institutions like INSAIT, it is very important not to get isolated. But how do you want to make sure that what you're doing here will also benefit Bulgaria and Europe and the AI scene here, and not just benefit those big tech companies? I don't think of INSAIT as an institute.

I think of it as a strategic base, for Europe even. So this has all kinds of tangible and intangible benefits long term. Digital rights advocates are more critical about such links between research in Europe and Big Tech abroad. I think it's a problem because even when external funding doesn't come with any strings attached. There's all these research institutions receiving big tech funding and are therefore less likely to be critical of big tech. Is there something in place to make sure that those kinds of collaborations and that their work actually benefits Europe? Well, if the collaboration takes place, it's because universities see a benefit.

And when we are capable of making research, we are also capable of innovating and then we should be capable of transferring this kind of innovation to our markets and our industry. Artificial intelligence is here to stay – and all experts we met on our journey across Europe agree that it will reshape the global balance of power. There will be two categories of companies and countries. The ones that have access to the best AI models and the ones that have not. But perhaps that depends on how you define success. While Europe may remain the underdog in terms of investment or innovation, the continent is leading the charge to define the limits of how AI can – and should – be used.

That's why the AI pioneers emphasize the importance of working together. If you want to go fast, then go alone. But if you want to go further, then go together. The Bulgarian researchers have big plans for the coming years. Here is our new building that we construct. Here we will install our big data infrastructure, a lot of technologies.

The computer scientist hopes the digital twin of her city she’s building will only be a beginning – and that it will inspire others. We want to scale the project, to cover for example the whole city and why not to cooperate with other cities in Europe and all over the world. AI is on the rise.

Now, Europe’s challenge is to scale up its efforts without scaling back protections for the people down below.

2023-08-20 01:07

Show Video

Other news