AI advancing on a broad front

Artificial intelligence is a hot topic at the moment, boiling hot, and the target for investment around the world, in Europe and in Sweden. AI has taken the step out from the research laboratory and we are experiencing the second generation of AI systems. The third generation will place people in the centre.

Jonas Unger in the AR laboratory, Campus Norrköping Thor Balkhed

The Center for Augmented Intelligence, CAI, has just been created at Linköping University, led by visualisation researcher Jonas Unger. This is a response to the growth of the third generation of AI systems, intelligent decision-support systems that come to the aid of our human brains when the amount of information to be processed becomes too great. The centre will work on the interface between people and AI systems, where the latter are to be tools that augment human intelligence.

Three generations

The first generation of AI research arose during the 1980s and 1990s. It dealt with computer science and logic, and Erik Sandewall, professor emeritus at LiU, was one of the trail-blazers. Machine learning characterised the second generation of AI research, in which AI systems learn from the huge amounts of data with the aid of neural networks, deep learning and other techniques. This technology now finds large-scale application by, for example, Google and Facebook.

The third generation will now see people introduced into the AI systems. One example is the research carried out by LiU-professor Patrick Doherty into delegation processes, in which autonomous drones, boats and helicopters collaborate with each other and with people to carry out rescue missions. A successful demonstration for sea rescue outside Västervik has recently been carried out within the framework of WASP – the Wallenberg AI Autonomous Systems and Software Program.

“If we take each AI system that we already have and add a human aspect, we create a much better system”, says Anders Ynnerman, professor at LiU, director of Visualization Center C, and head of research for Visual Sweden, a Vinnova Vinnväxt programme of which the Center for Augmented Intelligence is part.

Intelligence in our pockets

“We have become accustomed to having all information available in our pockets: in the future, we will have intelligence in our pockets. We must think about how this will affect us”, says Anders Ynnerman.

Together with Jonas Unger, he sees a future that contains cognitive companions, humanoid bots, maybe in the form of holograms, with whom we can hold a conversation when faced with important decisions. But we’re not there yet.

“Large data-driven systems are available that act as rapid encyclopaedias, but these have limited understanding of the context. We have achieved a great deal when it comes to sensory systems, the way in which the bots use sight, hearing and other senses. But work has just begun on the reasoning layer, which is the layer with higher cognitive functions and logical thought.”

In other words, AI can already help us solve many problems, but not all.

Affects all parts of society

Fredrik Heintz is another AI researcher at LiU. He is also director of the graduate school within WASP and member of the EU Commission high-level expert group in AI. He wants to see an even broader AI initiative across Europe, in order to truly place people at the centre of AI development.
“The public sector is one extremely important application area since welfare states are no longer able to deliver the level of service that the citizens expect. The large-scale coordination of healthcare and medical care, for example, is too difficult for us. We are not able to discover and keep track of patterns in enormous amounts of information, while AI has very effective methods for this”, he says.

He believes that AI will affect all parts of society, and that investment within the social sciences and humanities is needed. “Even if investment in AI is high at the moment, significantly larger and broader initiatives are needed”, he says.

Anders Ynnerman agrees:
“There are ethical, legal and societal aspects of AI, and it’s high time that we got to grips with these”, he says, and gives as an example the autonomous cars that must master ethics and morals. An autonomous car may be faced with making an ethics-based decision, such as: should the car swerve to avoid a dog that runs across the road if this involves risk to the passengers in the car?

Fredrik Heintz and Anders Ynnerman agree that LiU has a strong position in AI research. In addition to the research within scientific visualisation, and within artificial intelligence and integrated computer systems described above, Professor Michael Felsberg is carrying out advanced research into computer vision. This was recently recognised by Vinnova as Sweden’s leading AI research environment. Another field that LiU leads is the Analytic Imaging Diagnostic Arena AIDA, which is a national arena for research and innovation for AI within image analysis, where researchers, industrialists and medical personnel collaborate.

“We have several attractive environments at LiU and must recruit more researchers. But at the same time there is a global dearth of expertise in AI”, concludes Anders Ynnerman.

Translated by George Farrants

Contact

More AI news

Research

Latest news from LiU