«Intercultural competence creates trust and trust is a key element for a successful business relationship»

Marco Pellin, Senior Manager

News

Consenec Impuls:

A Closer Look at Artificial Intelligence

Artificial intelligence systems have become a part of our everyday lives. What will the next breakthroughs be? How reliable and trustworthy is the technology? Dr. Andreas Krause, Professor of Computer Science at ETH Zurich and member of the United Nations AI advisory body, discussed his views on the subject at the latest Consenec Impuls event in Dättwil.

The concept of artificial intelligence (AI) has long fascinated the human imagination. Already in 1950, Alan Turing explored the question as to whether machines could be taught how to think. And his Turing Test is still used to assess how well computers and software programs can simulate human behaviors. The term “artificial intelligence” itself was coined by John McCarthy in 1956.

New Consenec CEO Volker Stephan welcoming guests to the first “Consenec Impuls” under his leadership.

Fluctuating interest in AI
In the decades after the 1950s, AI saw a period of remarkable innovation, with developments like reliable language translation tools and chess-playing systems capable of defeating grandmasters. However, there have also been phases called “AI Winter,” during which interest and funding in the technology plummeted.

Professor Andreas Krause, who has various roles at ETH Zurich, including Chair of the AI Center …

AI is here to stay
This changed radically when ChatGPT was launched two years ago. Since then, the whole world has been talking about AI and there’s a general awareness of the technology in society. In academic circles, the introduction of the visual database ImageNet in 2012 is considered the breakthrough moment for AI. And this year, two physicists were awarded the Nobel Prize for their groundbreaking work on neural networks and machine learning – the foundation of many modern AI applications.

… drew a large audience to Dättwil with his talk on AI.

From specific to general
“In contrast to earlier, niche applications of machine learning like chess or language translation, we’re now seeing the advent of what could be called general artificial intelligence,” is how Dr. Andreas Krause framed the developments last Tuesday in Dättwil. He explained that these so-called “ foundation models” are large machine learning models that can be adapted for use in numerous applications – ChatGPT being one example.

Is it trustworthy?
In the ImageNet project mentioned above, AI now attains results that are as good, if not better, than those of humans – who sometimes also fail to correctly interpret the subject of blurred images. But, Krause said, caution is advised when using AI systems like ChatGPT: “When asked to answer factual questions, the chatbot doesn’t access information stored in databases. Rather, it operates on the basis of probability.” He then showed the audience how ChatGPT answered a question about the name of the first dog to land on the moon. Although “Laika” was indeed the first dog to be sent to outer space, she didn’t make it to the moon – nor have any of her fellow canines.

Uncertain future
Concerning further developments in AI, Dr. Andreas Krause was unwilling to go out on a limb and make predictions. He said he isn’t ready to say when he thinks autonomous vehicles will be a thing of reality, or whether something like superintelligence is truly possible.

Current applications in medicine and chemistry
He does, however, identify great potential for AI in the fields of medicine and chemistry. As examples, he cited the NCCR Catalysis, where he is collaborating, as well as the Meditron project (information in German) at ETH Lausanne and the fact that this year’s Nobel Prize in Chemistry is inherently related to AI: two chemists were awarded one-half of the 2024 prize for their development of an AI model that predicts protein structures.

Advances in AI thanks to enhanced computing power
Aside from potential applications, one of the greatest drivers of AI is simply the fact that computers are becoming more and more powerful. One such powerful machine is the new ETH Zurich Alps supercomputer that was commissioned last September at the Swiss National Supercomputing Centre in Lugano – which is also operated with solutions (information in German) developed at ABB.