Dusseldorf Artificial intelligence (AI) will change the world. It will facilitate research, support creative processes and automate work steps in factories. The chatbot ChatGPT from the US start-up OpenAI, in which Microsoft intends to invest billions of dollars in the next few years, is currently the focus of particular attention. The technology behind ChatGPT is as important as the invention of the Internet, said Microsoft co-founder Bill Gates in an interview with the Handelsblatt a few weeks ago.
Hundreds of start-ups are trying to develop new business models based on AI in Silicon Valley alone. Many of them are still at the beginning. But the goals of companies like OpenAI are big: they want to create computer programs that can learn almost any task.
Is this realistic? Is this evolution exponential, as some tech visionaries believe? What does all this mean for Europe? This is what top-class experts are discussing at the “Fireside Chat” of the Handelsblatt Media Group in Düsseldorf: Jonas Andrulis, founder and CEO of Aleph Alpha, and Tina Klüwer, director of the Artificial Intelligence Entrepreneurship Center (KIEZ) in Berlin.
Aleph Alpha is developing a European alternative to ChatGPT. KIEZ is a model project of the Berlin universities and supports science-related companies, especially promising AI companies. The so-called accelerator offers start-ups infrastructure, such as office space, access to job portals and mentoring from AI experts.
The group, which was attended by around 100 business representatives, was moderated by Handelsblatt Editor-in-Chief Sebastian Matthes. Those were the most important theses.
Thesis 1: AI will trigger an industrial revolution
There was agreement on the panel right from the start: ChatGPT will change almost every industry from the ground up. “We’re in the midst of an industrial revolution, and it’s the fastest yet,” says Andrulis.
According to the founder, all information-based value-added processes would be redesigned. AI can already compose and paint, write journalistic texts, solve exams and generate text modules with legal arguments.
>> Read here: In these areas, AI can relieve us of the creative work
KIEZ Director Klüwer is also convinced of the potential of AI and adds: “AI is the spearhead of digitization.” In recent years, more and more business and work areas have made the leap into the digital world.
As a result, digital data emerged over time in contrast to paper files, which now form the basis for the next step in digital transformation, cognitive processing using AI, says Klüwer. With the help of this data, the AI can learn to take over parts of the tasks that previously only humans could solve.
Numbers show how this key technology could continue. Worldwide sales of enterprise applications in the field of AI are estimated at around 31.2 billion US dollars in 2025. According to calculations by the market research institute Omdia, which focuses on technology, it will be almost eight billion US dollars in Europe in 2025.
Thesis 2: So far, people have adapted to the needs of the machine, now it’s the other way around
ChatGPT is based on a large language model, a deep learning system with many hundreds of billions of parameters that understands almost any form of language, and can often answer correctly – while completing tasks or passing on knowledge.
The chatbot is particularly powerful and versatile because a large dataset was used for training, in which people painstakingly noted their own preferences, says Andrulis.
The language model optimizes itself on the basis of this data set with human inputs. Whereas people used to have to learn the programming language of the computer, today it is the machine that “understands human language in all its complexity”.
Thesis 3: The greatest surges in innovation occur in the fields of medicine and sustainability
According to Klüwer, the biggest innovation boosts through AI are currently occurring in medical technology, among other things. AI in medicine is already in use. Deepmind, for example, the subsidiary of Google parent company Alphabet, uses a platform to help predict protein structures. The structure of a protein determines its function in the cell. If this structure is known, drugs can be developed more precisely.
Andrea Wasmuth, Managing Director of the Handelsblatt Media Group, and Editor-in-Chief Sebastian Matthes will open the evening.
(Photo: Max Brugger for Handelsblatt)
New are AI-based processes that create new proteins. To do this, researchers fed a speech generator with amino acids, the building blocks of proteins, and with information about their function. It is true that only a fraction of the proteins produced were able to kill bacteria like the real proteins. Nevertheless, it is “a quantum leap,” says Klüwer. In this way, too, new medicines can be developed.
In the area of sustainability, the greatest innovation pushes come from the development of technologies that remove CO2 from the air and process it as a raw material, says the AI expert, for example in the production of building materials, kerosene, cosmetics and food.
Thesis 4: The infrastructure of the digital world of the future is currently being created
The key to the AI infrastructure are programming interfaces, so-called “Application Programming Interfaces (API). In addition, all companies can use the OpenAI language models. Klüwer explains it using the example of typical customer service tasks. For example, if the employee receives a complaint email, they do not process it manually, but automatically via the OpenAI language model. “It’s much more efficient,” says Klüwer.
Until now, companies have built such AI functions themselves. But the big models are more powerful and better. The problem: As soon as German companies only build on the infrastructure of American or Chinese companies, “a part of the added value is lost to the big service providers, just like with cloud computing,” she says.
Thesis 5: In the field of AI, Europe has a fair chance, but not for much longer
Nevertheless, Klüwer is convinced that major tech players can emerge in Europe. “Europe is not starting from scratch,” she says. “We have a strong research landscape and creative minds at universities who want to build companies with innovative business models.” The decisive factor here is to transfer the research results into business models. “Germany has some catching up to do here.”
>> Read here: This is how more start-up culture should develop at German universities
If Germany succeeds in accelerating innovation in business and administration, “we have a better chance of winning the race than with cloud computing,” adds Andrulis.
Despite fundamental reluctance, investors have recently invested heavily in the generative AI sector, i.e. in systems such as ChatGPT that produce texts, images or program code. These investments took place primarily in the USA. In Germany and Europe, concern about the upcoming AI regulation is inhibiting some innovation projects, says Andrulis.
Thesis 6: AI does not automate manual jobs, but starts with knowledge workers
For a long time, there was a fear that AI would put thousands of people out of work. Today we know: AI is not advancing fast enough to take over all the jobs for which human labor is not available.
Contrary to the assumption, which is also widespread, that machines would first automate manual activities, Andrulis says, the revolution begins with knowledge workers. “It worries a lot of people.”
>> Read here: Which jobs artificial intelligence really threatens
Klüwer calls for a stronger focus on the opportunities offered by technology: if more work is done by machines, human work will have more quality value. AI is a “tool that supports human thinking”.
Andrulis goes one step further: “People are just machines too, biological machines.” In the long term, there are no insurmountable limits for AI.
Thesis 7: A key qualification of the future is adaptability and creativity
This makes the question all the more important: What qualifications will be important for people in the future? According to the panelists, there are essentially four skills that are deeply human and difficult for machines to learn: creativity, intellectual curiosity, the ability to adapt and the ability to solve new problems.
According to the panelists, there are essentially four skills that are deeply human and difficult for machines to learn: creativity, intellectual curiosity, the ability to adapt and the ability to solve new problems.
(Photo: Max Brugger for Handelsblatt)
In journalism, the trend is clear: Summaries of existing content will soon be taken over by machines. But an AI won’t be able to ask questions that nobody has asked before and find information that hasn’t been reported anywhere, says Klüwer. “It needs people here.”
The evening ended with a discussion among the audience. A professor asked how she could tell whether seminar papers were being written by real students or by machines – and whether they were still a useful means of assessing performance. “There are no easy answers,” says Klüwer. The key is the cooperation between man and machine in business as well as in science.
The discussion showed that Europe has good prerequisites for positioning itself as a leading technology hotspot in the field of AI. The implementation requires incentives for founding in universities and companies as well as a suitable political framework to build the AI infrastructure of the digital world of the future.
More: Six industries that could be transformed by AI