It’s a whole new form of romance that Feiyu Xu is talking about. “It’s easier to fall in love with a robot today than it was a year ago.” Because artificial intelligence (AI), says the entrepreneur on stage at the SZ economic summit in Berlin, can now imitate human behavior really well. Chat-GPT and other so-called generative AI models write texts, paint pictures, and have long conversations with their users. Chatbots, image generation programs and digital assistants hold a mirror up to people. But be careful, says Xu: generative AI can act like an intelligent being. “But she doesn’t have real emotions, she can only simulate them.” So it would be false love.
Xu is not a couples therapist, but rather an AI manager. The computational linguist was responsible for AI at the German Research Center for Artificial Intelligence for a long time, and later at Lenovo and SAP. On the podium she will discuss with the philosopher Markus Gabriel, professor at the University of Bonn. You’re talking about the technological emergency that the chatbot Chat-GPT triggered almost exactly a year ago.
Since then, humanity has felt that it is no longer alone among intelligent life forms. That’s a wrong feeling – even with AI, machines have no will and don’t even understand the world as well as small children. But the excitement that the so-called large language models of the tech companies have caused is enormous. No entrepreneur wants to be an entrepreneur without an AI strategy. AI talks to the lonely for hours, sets up emails and creates paintings that you can hang on your wall without shame. Artificial intelligence is therefore the topic of several rounds of discussions at the SZ economic summit.
Before Xu and Gabriel took the stage, Tina Klüwer from the Berlin Artificial Intelligence Entrepreneurship Center talked about that day a year ago when she spoke to Chat-GPT for the first time: “For me as a computational linguist, this chat effect was crazy.”
A new AI program could be tailored to individual industries and companies
This madness in turn motivated Feiyu Xu to leave SAP and start something new. With the start-up Nyonic she wants to enter the market for large language models from Berlin. She wants to sell artificial intelligence to companies. During their training, the AIs should absorb specialized knowledge of an industry as well as the internal knowledge and data of the respective company – the “industry language,” as Xu says. Such a program should then be a kind of good spirit for employees: looking for information, answering questions – an assistant who never leaves work.
Sergej Epp, head of security at the cybersecurity company Palo Alto, which protects tens of thousands of customers from hacker attacks, previously reported that AI can do more than chat. Millions of attacks have to be detected and prevented every day. “That doesn’t work without artificial intelligence.” Only with it could digital defenders like him build and scale their defenses efficiently. The attackers would do it too.
According to Epp, the biggest danger in Germany is probably not hackers, but rather “if we develop German fear again with AI.” 42 percent of respondents give one Opinion poll of the Association of the Internet Industry (Eco), they considered AI to be harmful to humanity. Only 27 percent believe that the benefits of the technology outweigh the benefits.
The uncertainty is fueled by Silicon Valley itself, where the best-known AI programs such as Chat-GPT or Bard from Google come from. Some entrepreneurs even warn of world domination or apocalypse by a “superintelligence.” The panic group also includes OpenAI boss Sam Altman, who is responsible for Chat-GPT, and Tesla boss Elon Musk.
AI tools for offices are not free
In addition to Tina Klüwer, Marianne Janik, Managing Director of Microsoft Germany, is also on the podium with Epp. It doesn’t have to justify an impending takeover by AI robots, but rather the $30 per month that Microsoft wants from customers who want to use new AI capabilities. As a major investor in OpenAI, Microsoft has privileged access to Chat-GPT. Janik says: “It’s not just Powerpoint, but also Excel and Outlook, for example.” The whole package for office workers, and “that just costs computing power”. Every company can test whether AI is worthwhile for them.
“Chat-GPT and co: chat machines, revolution or danger?” Moderator Bastian Brinkmann, deputy head of the SZ economics department, discussed this question with Sergej Epp (Palo Alto Networks), Marianne Janik (Microsoft), Tina Klüwer (Artificial Intelligence Entrepreneurship Center) and Jaroslaw Kutylowski (DeepL, from left to right)
(Photo: Johannes Simon/Johannes Simon)
Jaroslaw Kutylowski sits two chairs next to Janik. He doesn’t need AI to free himself from the yoke of Power Point presentations. With the translation tool DeepL, which he helped build, he is something of a silent hero of the German AI scene. DeepL is better than the translation service from the mega-corporation Google. Kutylowski says that he delegates the boring parts of being a programmer to a so-called AI co-pilot: “When I work on the program code, it helps. Not for the big idea, but for the smaller mechanical steps.” However, the user doesn’t even notice that there is a lot more AI in DeepL. It just works well.
Such a pragmatic approach grounds the AI debate this afternoon. Philosopher Gabriel Markus has also dealt with the dystopias of the apocalypticists. But he says: “The danger at the moment is that AI is too stupid.” That so many people use technology aboveestimated, could still be a problem. Markus also says: “I would put the IQ of the AI programs at zero.” So they are not only far from human intelligence. They didn’t even play in the same league.
And with the feelings that artificial intelligence won’t be able to handle that in the foreseeable future anyway, says Markus’ conversation partner Feiyu Xu: “An emotional escalation in customer service – that’s better left to people.” As long as rage and reconciliation remain in human hands, everything is fine.