AI for Switzerland—and the Rest of the World

Artificial intelligence is taking over our daily lives at an almost avalanche-like pace, whether in business or in our personal lives. There is hardly any area left that isn’t forced to deal with it. «AI is the electricity of the 21st century,» said Andrea Grisard, president of the event organizer SQS, in her opening remarks at the Hotel Schweizerhof in Bern. Needless to say, AI does not stop at SQS, a leading conformity assessment body: For decades, it has been assessing management systems, products, processes, and services—and more recently, it has also incorporated AI into these assessments. Not least, ISO/IEC 42001 is a new standard for AI governance. Andrea Grisard also emphasized the importance of certifications for securing market access.

Apertus – a Swiss LLM

In his keynote address, Prof. Dr. Thomas Schulthess of the Swiss National Supercomputing Centre provided interesting insights into the development of AI «Made in Switzerland.» With the ALPS supercomputer and the Apertus language model, Switzerland can not only apply AI but also develop it independently. Unlike other large language models (LLMs) such as GPT, Claude, Mistral, etc., Apertus is available entirely as open source. The source code for training processes and the training data itself are also disclosed. With this maximum level of transparency, Apertus meets all EU compliance requirements. Companies can use Apertus to develop their own language models and thereby contribute to further specializing the model. Schulthess emphasized that collaboration with the business community is particularly crucial. An important goal, he said, must be to enable SMEs to integrate AI into their business areas and systems. That is why strong partnerships are necessary—not least to spread the high costs across multiple parties.

Regulating AI: Yes, but how?

Politics then came into play during the subsequent panel discussion, particularly in connection with the question: How can AI be regulated? Alongside Thomas Schulthess and Martin Saladin, head of SECO’s Investment Promotion Division, two National Council members—Benjamin Fischer (SVP) and Gerhard Andrey (Greens)—answered questions from moderator Alex Gertschen. Benjamin Fischer warned against an «AI bubble» in which the major hyperscalers would essentially seek to dominate the market, following the motto «The winner takes it all.» Nevertheless, he is critical of overly strict regulation, as it could also hinder innovation. Gerhard Andrey, on the other hand, argued that innovation must take place not only in technology but also in governance. We should not «let everything that is technologically possible run wild,» and he described such «Elon Musk fantasies» as «toxic.» Prof. Dr. Thomas Schulthess noted a widespread lack of understanding of what is «really going on» with AI. He emphasized that values must be preserved, but not through defensive behavior. Furthermore, he advocated for a more active role for the private sector. Martin Saladin, on the other hand, said that not everything has to be governed by laws, but rather by the will and interests of the various stakeholders. So, is self-regulation a goal? In principle, yes, «but that also requires data,» according to Thomas Schulthess. After all, to train language models in ethics, one must also be allowed to work with «unethical» data. And what about the energy consumption of data centers? While in the U.S., for example, gigantic new data centers—some with their own power plants—are being built «just so kids can generate AI-generated cat videos and other data junk,» as Gerhard Andrey put it, priorities in Switzerland need to be set differently. Andrey therefore calls for data centers built here to be primarily dedicated to industrial purposes.

The «Human Loop» Remains Central

The subsequent workshops then focused on practical aspects of AI. For example, the Swiss Standards Association (SNV) provided information on regulatory developments related to AI infrastructure in Switzerland. A second workshop addressed challenges and potential solutions for organizations with a critical risk profile. Two workshops were dedicated to concrete implementation options in companies. For example, Endress+Hauser, represented by Jürgen Wirth and Thorsten Springmann, demonstrated how the company develops, tests, and implements its own AI agents. On the one hand, this makes it possible to solve complex tasks more quickly and automate processes. On the other hand, however, this does not result in job cuts but rather in the enhancement of roles, specifically in terms of accountability—decisions are still made by specialists—and governance.

The Interplay Between Science, Business, and Politics

This small but excellent conference demonstrated that artificial intelligence should not—and must not—be left solely to the major players. With ALPS and the Apertus language model, Switzerland can make a contribution to further developments that, while small in scale, allows it to play a pioneering role in terms of quality and transparency. And the fact that our country sometimes takes its time with regulations isn’t necessarily a disadvantage. After all, the EU’s AI Act could backfire on Europe, as there is a risk of slowing down its own technological development. And that, in turn, would be disastrous for small economies like Switzerland’s. This makes targeted collaboration between science, business, and politics all the more essential—that is certainly a key takeaway from the SQS Corporate Forum 2026.

Information about SQS: www.sqs.ch

This article originally appeared on m-q.ch - https://www.m-q.ch/de/ki-fuer-die-schweiz-und-den-rest-der-welt/