The next generation of artificial intelligence, fusing intuition and rational thinking: Research team at the University of Klagenfurt participates in “Cluster of Excellence”

The capabilities of artificially intelligent systems are still somewhat limited at present. This can be attributed in part to the fact that there are two streams of research in AI development and so far these have largely been pursued in parallel. The Cluster of Excellence titled “Bilateral Artificial Intelligence” aims to combine the strengths of both methods in bilateral artificial intelligence in order to substantially enhance the performance of AI systems. On the 6th of May 2024, the cluster was presented as one of two new projects by the Austrian Science Fund FWF. Clusters of Excellence are regarded as Austria’s beacons of basic research. 

There are essentially two streams in artificial intelligence research: Symbolic AI is based on logic, while sub-symbolic AI involves training artificial neural networks with vast quantities of data. Research has achieved considerable progress in both fields, albeit largely separately and asynchronously. In order to be able to move on from specialised AI systems to a “broad AI” with more advanced and more universal problem-solving capabilities, the consortium aims to combine both in the “Cluster of Excellence” on “Bilateral Artificial Intelligence”. “The broad AI should offer both an understanding of the world and comprehensive cognitive abilities. Broad AI should be able to master complex tasks effectively on the basis of its sensor perception, its previous experience, its learnt skills, its in-depth knowledge of the environment and its logical thinking, and it should be able to adapt to new tasks fairly automatically. Broad AI can justify its behaviour and share its knowledge with other intelligent agents,” states Gerhard Friedrich (Department of Artificial Intelligence and Cybersecurity at the University of Klagenfurt), and member of the Board of Directors. The consortium is led by Sepp Hochreiter (University of Linz). The University of Klagenfurt will be able to contribute its expertise in the field of logic-based AI systems in particular, and the practical application in terms of solving large and difficult problems.

The consortium cites AI-controlled traffic management systems as an example. At present, specialised AI systems are being developed and adapted by humans to solve a range of traffic control tasks in different regions. This AI can control traffic, prevent air pollution and time loss by optimising vehicle routes, adjust the phasing of traffic lights, regulate the speed of vehicles and guide emergency vehicles to their destination. By contrast, broad AI systems are not designed by humans to perform special tasks, but instead they are self-optimising and can adapt to changes autonomously. They possess a high degree of flexibility that allows them to automatically utilise new technical possibilities in the field of sensor technology and control. Having a sound understanding of the laws and causalities of the world facilitates these adaptive capabilities, improves the safety of AI systems and supports the justification of actions. New tasks can be learned through dialogue with humans. Broad AI can respond robustly to disruptions such as sensor errors, weather events or attempts to attack the infrastructure. It learns about traffic patterns through observation and can use and pass on this knowledge to new regions and cities.

Gerhard Friedrich explains: “Humans are highly flexible when it comes to adapting to new situations and solving new tasks. They can combine experience with basic knowledge and learn to solve tasks better and more efficiently. Broad AI will bring these characteristics to our computer systems.”

The other members of the board are Martina Seidl and Sepp Hochreiter (University of Linz), Christoph Lampert (IST Austria), Axel Polleres (Vienna University of Economics and Business), Robert Legenstein (Graz University of Technology), Agata Ciabattoni and Thomas Eiter (Vienna University of Technology).

By launching the excellence initiative excellent=austria in 2023, Austria has opened a new chapter in basic research: Five Clusters of Excellence are currently carrying out cooperative projects at eleven locations. The FWF and the participating research centres have committed an investment volume of 135 million euros. Now, a further two clusters can begin their work starting in 2024 thanks to funding. The new clusters were presented to the public on the 6th of May 2024.

Photo can be used free of charge for reporting on the project: