AI Is Unlocking New Possibilities for Machines and Manufacturing

House of AI is using artificial intelligence to develop more efficient and safer machines. Collaboration with companies also deepens the researchers’ expertise.

One of the three focus areas of House of AI, a research center combining AI research with industry expertise, focuses on manufacturing and autonomous machines. Professor Ville Kyrki, who leads this focus area, says that it is also the area where activity has grown the fastest.

“One reason for this is probably that Finland has a strong industrial tradition in machine manufacturing. In the past, we made paper machines; today, we are more focused on forestry machinery, mining equipment, port cranes, elevators, and many other types of machines and equipment.”

According to Kyrki, all these fields share a substantial scientific and technological foundation. At House of AI, researchers are studying how artificial intelligence can process information originating from these machines and control them in ways that make them more efficient and safer.

Finnish companies have also been at the forefront of earlier development trends in the field, including the Industrial Internet and the use of big data. As companies have been collecting operational data for a long time, various AI models now offer new ways of making use of it.

“Big data involves analytics, meaning that you can find information in the data. It is somewhat like statistics with a lot of added sophistication. AI, on the other hand, has made it possible to process data in more complex forms, such as human-written language, maintenance reports, or medical diagnoses.”

Another opportunity offered by AI is its ability to make recommendations and provide reasoning for them. Instead of simply searching data for, for example, the most common diseases or symptoms and their associated exposures, AI can be used to guide processes rather than merely analyze data.

Access to real-world data from companies also helps researchers develop their expertise.

“It is very interesting for researchers to be able to work with genuine, large-scale datasets. Current AI requires very large amounts of data, and this is often only available within companies. Through collaboration, we are able to gain a deeper understanding of how AI methods can be applied,” Kyrki explains.

The rapid spread of AI applications has also surprised researchers, at least to some extent, according to Kyrki. He says that technological breakthroughs usually happen through a series of small advances until their combined effect reaches a critical threshold.

AI researchers have been aware of these individual advances, but predicting when a breakthrough will occur is more difficult.

“For example, smartphone technology saw simultaneous advances in both processor capacity and touchscreen technology—and suddenly we had a computer in our pocket. Something similar happened with AI: a threshold was crossed after which entirely new possibilities emerged. This happened a few years ago with language models,” Kyrki explains.

The technology developed for language models can now be applied to more than just natural languages—for example, to programming languages, music, and images.

In Kyrki’s field, autonomous cars have long been predicted to take over the streets, but the major breakthrough has yet to materialize. Instead, there have been—and continue to be—many smaller advances. For example, the interface between humans and machines is undergoing a major transformation.

“In autonomous machines, physical AI is still a few steps behind, which is why it is being researched so intensively. However, we have already reached a point where, for example, you can give a robot tasks in natural language: you can instruct it verbally, and it will carry out the commands.”

Many Finnish companies have now recognized the significance of the breakthrough in physical AI. At present, researchers at Aalto House of AI are working together with companies to develop more capable, efficient, and safer machines, ranging from walking robots to forestry machinery.

Text: Heidi Hammarsten