In a new laboratory in Davos, scientists from all over the world are working to create algorithms with human intelligence. The Swiss Alpine city wants to become known as a research center for “neutral political” intelligence to counter the influence of China and the United States.
This content was published on February 2, 2023 – 09:00
Michele Andina (Video)
In a historic villa in the center of Davos, scientists are trying to understand the basis of human intelligence. They believe that brain decision-making is the key to developing artificial intelligence (AI) that works for people instead of governments or big interests. In this way, they hope to help solve the biggest problems of our century, such as climate change and disease.
Thanks to its status as a middle country with strong research capabilities, Switzerland is able to challenge the systems of China and the United States, which have used AI to control their models of power: control on the one hand and capital on the other. .
“Globally, we need our third AI research that does not operate as a commercial or state-owned enterprise,” says Davos mayor Philipp Wilhelm .
A center for neutral AI
Davos, known for hosting the World Economic Forum (WEF) every year, has long been home to famous research institutes. But until recently, AI was not a topic in the valley in the south of Switzerland: it has been associated with many big cities of Zurich, Lausanne and Lugano. That is until Pascal Kaufmann, a Zurich-based neuroscientist with a passion for language and psychology, decided to set up an international laboratory for the “human-level”. AI research, in Davos. Lab42 opens its doors at Villa Fontana in July 2022.
Both Kaufmann and Wilhelm believe that the Alpine resort, with its 11,000 residents, has what it takes to attract talent from around the world and become another Swiss AI center.
“Davos is an international science city, put in a very good situation,” says Kaufmann, “The clean air and the process are the best thanks to the WEF.”
Video: Davos Mayor Philipp Wilhelm explains how the mountain has been developed as a major research center.
Kaufmann has a self-described “friendly” relationship with Davos. His NGO Mindfire, which advocates for AI that predicts human intelligence, started its first step to break the rules of our brain in Davos in 2018, inviting experts from around the world. A few years later, when his team wanted to turn this project into a laboratory with physical space and a virtual community, the city welcomed them with open arms.
“The Davos approach is to explore the most important problems in the world,” said the city’s mayor Wilhelm. “Digitalization is one of them.”
But the network that Kaufmann and his colleagues want to build goes far beyond the Alps. Using online challenges in the form of tournaments or multi-level games that currently machine learning approaches and algorithms are not yet able to solve, the Lab42 team searches for the best minds in the work of AI around the world and connect them with each other. An example is the Abstraction of Reasoning Corpus (ARC), a challenge that French software engineer and AI researcher François Chollet, now working for Google, created in 2019. ARC is considered intelligence tests for algorithms and there are 1,000 different jobs, various. which requires the ability to understand problems that AI does not have today.
Participants whose algorithms solve the biggest tasks win cash prizes and are invited to stay in Davos and contribute to research in the laboratory. Around 100 scientists and collaborators have visited the lab since July, but Kaufmann says it’s not enough.
“To make progress in AI, we need hundreds of thousands of researchers to work together,” he said.
The lab, financed by donations, employs many researchers. Donors include many public institutions, such as the Swiss cantons and the city of Davos, as well as the manufacturer MaxonMotor and the banks UBS and GKB. There are other private sponsors, but Kaufmann did not reveal their names.
‘Human intelligence is not enough’
So far, the most AI-based artificial intelligence tools have managed to solve only 20% of the ARC tests. For this reason, Kaufmann and his team are asked to understand how the human brain works in order to improve AI in areas such as abstract cognition and reasoning.
“Human intelligence is not enough to solve the world’s problems, but that is where we have to start,” he said.
Lab42 recently launched an international competition called ARCathon II, with the hope of attracting more talented people to its circle. “We want to get to the point where we can create robots that can do some difficult tasks, like planting trees to fight climate change, or developing treatments for diseases that irremediable,” said Rolf Pfister, the laboratory’s director.
AI testing is not the only way to this end: the laboratory also has an essay competition, which ended in December, to collect opinions on the subject of artificial intelligence from experts and disciplines that are not directly connected to AI, such as philosophy, biology, and the arts. The third winner is the music student.
“It’s definitely an outside perspective that’s useful and provides new ideas,” says Pfister.
According to Pfister, most technology companies rely on the same methods, such as those used for chatbot ChatGPT, although they have limitations. Since its launch in November last year, this technology – capable of simulating and recording human-like interactions – continues to make waves, with technology entrepreneurs Elon Musk tweetingExternal links it is “pretty scary” and “not far from scary strong AI”. But Pfister believes that although ChatGPT provides good results, it has no understanding of the world and is therefore not reliable.
This type of understanding is at the heart of human-level and human-centric AI, says Kaufmann. “When we know the principle of artificial intelligence, Europe will finally be able to be effective in the use of human-level AI and compete with China and the United States, which mainly focus for optimizing deep learning and error forcing,” he said. Kaufmann believes Davos and Switzerland will provide a non-political environment for the development of responsible, inclusive and democratic AI-based technology.
Sophie-Charlotte Fischer of the Center for Security Studies (CSS) at the federal technology institute ETH Zurich also believes that Switzerland can play an important role in the field of AI. Fischer sees Switzerland as a valuable partner for international AI research and governance because it is the home of the European headquarters of the United Nations, one of the world’s leading countries. most in the world, and is neutral and not a member of the European. Union.
However, Fischer, whose research focuses on AI governance and technology competition between the US and China, also said that cooperation on AI is becoming increasingly difficult. growing because global competition is intensifying – as evidenced by the recent US export targeting of China’s semiconductor industry.
But Davos, while not overrun by tourists or WEF visitors during the winter, is still the only one around. Its mayor Wilhelm says that this is not a problem in the digital age, pointing out that the dependence on big cities has decreased while work-life balance is more important. Davos, with its research institute, skiing in the winter and hiking in the summer, offers a good life, he added.
But the lack of shelter, especially for local people and workers, is a concern for Wilhelm.
“We are working hard on the housing strategy to ensure that in the coming years there will be enough housing for families, accessible to people of different classes,” he said. Davos’ family policy and job prospects for the younger generation are close to the heart of the 33-year-old Wilhelm, a Social Democrat and one of the youngest leaders in the city’s history.
“We want our young people involved in science to be successful in Davos.”
Edited by Sabrina Weiss and Veronica DeVore
Follow JTI standards
More: SWI swissinfo.ch is accredited by the Journalism Trust Initiative