������

Artificial intelligence is already being harnessed for human learning, but in the near future there will be whole new learning experiences, which have not been possible before.

A straightforward way to make use of AI is to enhance the current learning material production process and improve the quality. Speech synthesizers, machine translations and generative models can help instructional designers, graphic designers and content writers in their tasks.

“Future learning solutions include completely new learning experiences, like totally personalized content, dialogue-based learning, various virtual reality applications, automated curriculum design and other applications where AI has an important role,” says Antti Keurulainen, the founder of and doctoral researcher at .

There is a long history of co-operation between Bitville and Aalto University, the coordinating institution of FCAI.

“Bitville started from a pilot project when we got sponsor money from Nokia to produce a CD-ROM of illustrative animations for a ‘signals and systems’ course at the Helsinki University of Technology [the predecessor of Aalto University]. This set of educational animations were used widely in the departments of electrical engineering and computer science by thousands of students for more than 10 years,” says Keurulainen.

The current co-operation with FCAI started a couple of years ago.

“It started to be clear that AI is going to play a major role in future digital learning solutions. I contacted people at FCAI and suggested co-operation, and started my own post-graduate studies as well,” explains Keurulainen.

Exploring the co-operation between AI and humans

FCAI and Bitville researchers are exploring how AI and humans can co-operate and how AI systems could help humans learn complex skills involving cognition and creativity.

“The thing that is exciting is trying to get the AI system to understand people deeply enough that they can be sensitive to the individual learning styles. The AI system might not know what knowledge you already have, so it needs to ask you questions, try to figure out what it is that you know already so it can design the perfect lesson for you,” says Andrew Howes, FCAI Visiting Professor from the University of Birmingham. Howes is a cognitive scientist who is specialized in human-computer interaction.

But AI systems do not just observe and interact to understand human behaviour. In the best case, they also understand the reasoning behind why a person is doing what they do.

“In the first two projects we studied the basic capabilities for an AI agent to classify behaviours and in limited settings adjust their assistive behaviour. In the third project we apply similar techniques to produce user models instead of user classifications. These models resemble real human users and teach AI to infer the parameters representing a wide range of individual characteristics. An advantage of this approach is very high efficiency. Where other approaches might need hundreds of interactions to learn about the user, ours can achieve the same with a few handfuls of trials. In this case, AI needs to select very carefully how to interact with a human to gain as much as possible information from the interactions,” says Keurulainen.

Global community of top researchers

Professor Andrew Howes has been part of of FCAI’s global community researchers since 2020.

”FCAI has very rapidly gained an international reputation for its research and applied research with companies. It´s done on the basis of both a deep understanding of machine learning and artificial intelligence, but also by establishing a network of international collaborations,” says Howes.

Antti Keurulainen emphasises that Aalto University has a long history of AI research and has produced world-leading research in the area of AI for decades.

“ they developed is just one example of that.  Because of this high-quality research and reputation, FCAI has been able to create a very interesting global community of top researchers around AI and related fields, and our co-operation project is a good example how we benefit from that as a small private company,” says Keurulainen.

The speed at which AI and technology evolve and develop is accelerating. This leads to a situation where the time gap between academic research and deployment in the industry is shortening.

“In extreme cases, when a research paper, together with code and even a working model is published, it could be taken into production use within weeks instead of years,” says Keurulainen.