The Nordic Ethical AI Landscape showcases Nordic companies and institutions who have taken a special focus on ethical AI and responsible use of data. The landscape will be updated on a regular basis and organizations can apply to become a part of it.
The landscape addresses the question of how Nordic businesses can best position themselves at the forefront of legislative, technical and ethical developments in order to become leading in AI & Data field. Surveys show that more and more businesses have ethics guidelines and principles but at large lack know-how of putting these into practice in their AI solutions.
The first version of the Nordic Ethical AI Landscape was published in December 2022, and included 24 companies, 4 consultancy firms and 12 institutions selected among a total of 200 applications.
In order to get a deeper understanding of the new project, we met with innovation advisers of the Nordic AI and Data program: Anna-Maija Sunnanmark and Olivia Rekman for a talk about the new project.
What is the purpose behind the Nordic Ethical AI Landscape?
“First of all, we want Nordic businesses to develop a competitive advantage in AI by focusing on ethical AI and responsible use of data”, Olivia explains.
“Apart from serving as a communication and branding tool for the Nordic businesses and institutions involved, the landscape can function as an information tool in sourcing new collaboration partners, potential customers, subcontractors and investment prospects as well as a knowledge repository”, she continues.
“Our aim is that the landscape can be used as a platform for matching companies and organizations within the Nordics but also globally, to help them on their ethical AI journey. In this way, we hope to propel ethical AI capabilities in Nordic companies which are needed to meet the Nordic Council of Ministers’ 2030 vision”.
What do you mean by ethical AI and why focus on ethical AI?
“AI is making a rapid entry into our lives. In a survey performed by PAGA, only 33% of consumers believed that they use technology that features AI. However, in reality, 77% of all consumers use an AI-powered service. Also, ethical AI encompasses the question of how AI will impact the context it is applied in. Ethical AI considers both positive and negative effects of the technology”, says Anna-Maija.
“The ethics of AI depend on the application of the technology. Similar AI-based technologies used for unlocking your phone can be used for other purposes such as warfare, recognition of individuals, or in social scoring systems”, she continues.
Why not just focus on functional AI, which abides by the regulatory framework in place?
“Making AI work for the well-being of people, animals and fauna and environment is a complex matter where a fully encompassing legal framework is not always in place. Parallels are drawn to the early era of the internet before it became more regulated with for instance GDPR”, says Olivia.
“When we speak of ethical AI, it also goes beyond issues that are not yet governed by law and might be difficult to fully hedge against with legislation. Developers and the data input are the key drivers which dictate how AI-algorithms will impact the context applied in”, she continues.
“There are many examples of statistical data input for AI-models that exacerbate already existing biases of our society. These might be true data but have unfortunate, and sometimes unforeseeable consequences in feeding into already existing structural problems and trends like discrimination against race, gender and origin. These type of biases and discrimination is only one of the many risks with AI. Therefore, there is a need to go beyond the legal framework to ensure preliminary damage control”, ads Anna-Maija.
Criteria for being selected to the Nordic Ethical Landscape
The Nordic Ethical Landscape will be updated regularly based on a range of criteria.
Criteria for startups
Data for AI
- Startups with a focus on ensuring data quality (labeling, sourcing, data privacy (via synthetic data or algorithmic / differential privacy)
- Startups focusing solely on data observability are not included
MLOps, Monitoring, and Observability
- Startups with a focus on providing fairness or bias evaluation for models pre-, post-, or in-training
- Some startups focus solely on ML monitoring in the traditional sense in that they monitor for failures of a more general kind. These are not included
AI Governance, Risk, and Compliance
- Startups offering risk analysis or governance tools that expand outputs of data science teams into a business or legal context
Targeted AI Solutions and Technologies
- Startups applying AI to solve an area of society that has historically been biased or unfair such as crime detection and recruiting
- Startups with a novel technology that fixes biased alternatives are also included. Examples include toxic content moderation technology and ethical facial recognition
Criteria for Institutions
Universities
- Included universities have multiple ongoing projects that specifically target ethics or bias reduction in AI / ML approaches
AI Research Institutes
- Similarly, considered AI research institutes may have a research category under AI / ML that pertains to ethics or responsible AI or have ongoing projects that directly relate to ethical AI
Criteria for consulting firms
- Firms providing AI strategy with a specialty or case history of implementing responsible, ethical solutions
- Firms that focus solely on AI strategy or general consulting are not included
Project partners
The Nordic Ethical AI Landscape is developed in collaboration with Ethical AI Database (EAIDB) and Ethical AI Governance Group (EAIGG). The two partners have constructed the landscape and selected the companies.
If you want to become a part of the landscape, you can apply here.
