In November 2022, Silo AI published the Nordic state of AI 2022, its second annual report providing an overview of what’s happening with artificial intelligence (AI) in Denmark, Finland, Norway and Sweden.
Silo AI provides professional services in AI from its nine offices in Denmark, Finland and Sweden. Since it was founded in 2017, it has grown into one of Europe’s largest private AI labs, with 240 experts including 120 PhD-level AI specialists.
According to chief technology officer Niko Vuokko, the Nordics are on the right track with data and AI, but a lot remains to be done to unlock the true potential of machine learning technologies.
“The industry-leading products will be elevated by AI, one way or another,” he told Computer Weekly. “As can be seen in the Nordic state of AI report, a large majority of Nordic companies already invest in product development to add AI features to their products. However, what’s lacking is a systematic way to bring AI technologies into production, to benefit the customers that use the products and services of the Nordic companies. This requires not only new types of expertise, but also new ways of organising AI-driven product development.”
One of the challenges faced by Nordic countries is gaining critical mass. Because of the relatively small populations, it’s hard to build up a group of experts and a set of best practices to share. The region has less than 2% of the worldwide AI talent pool, and Nordic countries have made a conscious decision to put more emphasis on the quality of the workforce and less on quantity. They have also put effort into attracting AI talent from abroad – and this has paid off, creating a positive talent migration trend.
All the Nordic countries face challenges in building products with natural language features for smaller language groups. Because none of the languages are big world languages, there are no large sets of native language data to train AI models.
The Nordics are generally early adopters of technology – and AI is no exception. More than 25% of the Nordic companies are already investing at least 20% of their research and development budget in AI projects. Moreover, the Nordic countries are planning to get ahead – or at least keep up with other industrial nations. Each of the four countries have at least one top-ranking AI-related educational institution – and private investment in AI has more than doubled in the region since 2021.
Nordic companies apply AI mostly to their core offerings – to improve the quality of both their products and services, and the features and usability of their products. Using AI in supporting functions is a secondary priority, but many companies have already made forays into this area as well.
Almost all surveyed organisations plan to continue investing in AI, and more than half expect to see new technologies in the next six months. “As the report shows, AI maturity is high in the Nordics, and it’s great to see that companies plan to invest even in these difficult times,” said Vuokko.
“However, in addition to forming new AI use cases, many companies are now rethinking their approach to AI. Instead of a five-year investment cycle, they’re now planning for a multi-decade shift that may truly change what they offer to their customers and how. This focus on new, long-term investments is likely to widen the gap between those who already know how to take advantage of production-grade AI technologies and those who get stuck with experimenting.”
Danish AI
Denmark’s national AI strategy emphasises AI ethics and building homegrown solutions. The strategy focuses on improving public services through AI, encouraging adoption among businesses and supporting a dynamic research community. One good sign of enthusiasm in Denmark is that private investments in AI have increased in the past few years.
The biggest challenge for Danish companies is finding enough technical people to run the projects. This problem manifests itself most often when it comes to turning a proof-of-concept project into an ongoing part of the business – or into a product, in cases where companies are in the business of selling technology.
Another big challenge for Denmark is keeping up with the computational needs of AI algorithms, which are demanding more computing power every year. There is also a lack of training data specific to the needs of the population. Part of the problem is getting good data without compromising data privacy – something that is particularly challenging with medical and demographic data. Another part of the problem with not having enough data is that the population of Danish speakers is too small to generate as much data as some of the other industrialised nations.
Denmark is particularly strong in the hearing systems sectors. The country is a leading exporter of hearing aids, and AI is playing an increasingly important role in this domain, with machine learning becoming essential for the advanced audio modelling used in hearing devices. Pharmaceutical, media and the public sector are also early users of AI in Denmark.
Finland’s strategy
Finland published its national AI strategy in 2017 – it was one of the first countries in the world to do so. The strategy is updated regularly, with an increasing focus on supporting implementation and deployment rather than abstract ideas and guidelines.
Finnish AI research runs primarily along three different dimensions. The first is to optimise the performance of AI algorithms to head off the problem where computational requirements get too far ahead of what hardware can deliver. As a small country, Finland is particularly sensitive to the increasing costs of computational power – even though they house what is currently Europe’s most powerful supercomputer, LUMI.
The second dimension is trustworthy AI. Ethics and values are important to Finland, as they are in all other Nordic countries. Research in trustworthy AI aims to overcome the complex ethical challenges inherent to AI.
The third dimension is improving human-machine interactions. The aim of this research is to reinforce the role of AI as a tool that supports humans.
Over in Norway
Norway published its National strategy for artificial intelligence in January 2020. The country hopes to support key industries with AI. This includes healthcare, maritime, energy and public administration.
It launched an AI-specific regulatory sandbox project in March 2021, which will help companies ensure that their AI offerings comply with GDPR by providing them with an environment where they can run tests.
Of the four Nordic countries surveyed by Silo AI, Norway has the smallest percentage of companies responding that they are already using AI. It’s worth noting that Norway is also the only country of the four where Silo AI does not have an office.
For the good of the Swedes
Sweden launched its national AI strategy in 2018. The government aims to use AI to increase its citizens’ welfare and gain a competitive advantage.
Sweden has a very favourable political climate for AI. The country funds research and development through different agencies, most notably Vinnova, which by April 2021, was funding 256 AI-related projects.
Part of the funding is directed at AI Sweden, the national centre for applied AI research and innovation. The mission of AI Sweden is to accelerate the use of AI, colocate competence and run projects of national interest, in collaboration with nearly 70 partners.