As part of Bavaria's High-Tech Agenda, Würzburg is a hub in a state-wide artificial intelligence (AI) network. With the planned CAIRO AI centre, FHWS is focusing here on an area which has long appeared to be the preserve of science fiction films: a humanoid AI.
A household robot moves through the home. It discovers laundry on the floor and puts the clean laundry away tidily in the wardrobe, the rest in the laundry basket. It then empties the dishwasher – before 6 pm because that’s when its owner wants to watch the sports show in peace. A futuristic scenario, because even though there are technological household helpers like robot vacuum cleaners, the help ends with vacuuming and mopping. The robot is still lacking one thing, namely: artificial intelligence which can think and act like a human. Such an AI could initially be used in the production industry above all, but could also later be used in all kinds of other fields. FHWS has set itself the goals of getting closer to a humanoid AI and has established the “Competence Center Artificial Intelligence and Robotics” (CAIRO) to this end.
CAIRO as part of the Bavarian AI network
FHWS had the idea for an AI centre in Würzburg a few years ago when the state government declared Bavaria a digitalisation state in 2017. But the plans for the CERI robotics centre in Schweinfurt, which had been in the pipeline for a long time, needed fleshed out first. A new idea then sprang up during the development of CERI, says FHWS President Robert Grebner: “It was clear that it was all about intelligent robotics, and AI is required to make robots intelligent. We therefore concluded that we also need an AI centre.” A financing opportunity opened up two years later with Bavaria's High-Tech Agenda. Within this framework, the state provided 360 million euros for 100 new AI professorships in order to establish an AI network across Bavaria. Both FHWS and the Julius Maximilian University (JMU) were able to prevail in a competition to find the best AI concepts, meaning that Würzburg was declared a hub for data science. Alongside three other hubs and the centre in Munich, Würzburg therefore takes on a key role in the AI network.
FHWS wants to supplement the three CAIRO professorships financed by the state, which are currently being advertised, with two professorships, for the time being, which are financed in-house. This is the foundation for the establishment of an AI centre which should act not only as a research institution, but also as a competence centre and contact for other higher education institutions and companies throughout Bavaria. There will also be an “Artificial intelligence” master’s programme for the first time in Germany from the 2021 winter semester, through which students studying computer science can specialise in AI.
The focus: strong AI
Both the teaching and the research work will deal with current AI developments and technologies. But FHWS wants to make its mark with a particularly innovative focus: so-called “strong AI”, also known as superintelligence. This is characterised by intellectual capabilities which equal or even exceed those of a human being. It can perceive stimuli in a sophisticated manner and work cognitively, act flexibly and creatively and is therefore self-aware. Its opposite is “weak AI” which can only solve the one highly complex task for which it has been specifically programmed. Since it does not develop a deeper understanding of the actual problem, weak AI fails as soon as a detail differs from the prescribed process. Although developments in virtual assistants such as Alexa or Siri, for example, often give the impression of intelligent software, research is still far away from strong AI. Robotics programme director Prof. Jean Meyer, who developed the CAIRO concept together with President Grebner, also confirms this: “All of the innovations which have been reported as ground-breaking intelligence in recent years are, rationally speaking, weak AI. Strong AI is a different thing entirely and is far beyond today's technical expertise.”
In order to make strong AI possible at all, it must be equipped with three areas of intelligence: perception, reasoning and action. With CAIRO, the three disciplines of AI should now be brought together under one roof, whereby each professorship will have one of the three areas of intelligence as a field of research. Because AI can only go through a comprehensive problem-solving process when all three areas mesh together: perceiving a signal, recognising the problem, creating a solution, implementing this in an action and drawing new insights from this.
The robomotive ecosystem
Grebner also considers it necessary for the research to remain realistic as well as having a focus on strong AI. “AI cannot deal only with itself, otherwise even such strong AI would be useless. It needs interfaces with the outside world in order to bring AI into the application,” the FHWS President stresses. “That’s why we are considering introducing an AI application professorship in each of our ten faculties which combines AI expertise and knowledge with the requirements in the respective field.”
CAIRO will also work closely with the CERI robotics centre in Schweinfurt. A “robomotive ecosystem” in which intelligent robots can be tailor-made should result: CAIRO develops AI solutions which CERI then implements and tests in robots. Robot manufacturers adopt these solutions which are used by companies in various fields. The resulting insights will be taken on in the appropriate faculty and passed back to CAIRO through the AI interface.
Germany owns the centre field
Research activity on AI has not only increased in Germany. In the knowledge that AI will play an increasingly important role in the national economy, every country is trying to lead the field in the competition for technological leadership. The number of AI start-ups in a country, among other things, is considered to be an indicator. For Germany, the Bundesverband Deutsche Startups (German Start-ups Association) currently identifies 275 AI start-ups. An assessment by the Institut der Deutschen Wirtschaft (German Economic Institute) in 2019, in which the number of AI start-ups is used as the measured variable and is viewed in relation to the population, shows that Germany is not being left behind here, as is often assumed. Here, Germany ranks even above China and Japan in tenth place, and thus in the middle of the top 15 countries. First place goes to Israel, followed by Finland and Sweden. However, such statistics only relate to weak AI, Meyer explains: “Strong AI is seldom a topic in current research. That makes it difficult to draw comparisons between the various countries.”
Strong AI – a realistic future scenario?
FHWS sees that as a challenge – even if the question of whether strong AI is possible at all is disputed. In a study by the US marketing research company Emerj in 2019, 45% of participating AI researchers still expected the development of a strong AI before 2060, while 21% considered it to be improbable that strong AI will ever exist. Grebner, being a computer scientist himself, responds to the doubters: “Anyone who doesn't believe that strong AI is possible won't be able to find the concept for how it could work. But our goal is precisely this discovery. Whether it will work in ten or fifteen years, or even never – many great developments which benefit society and individuals will be made on the road to it.”
