Go to content

Adoption of Ethical and Responsible AI in the Nordics

Ikon3.png

State of AI in the Nordics

In 2022, over 100 senior executives of major enterprises from Denmark, Finland, Sweden, and Norway were surveyed on their AI maturity. This research showed that companies in this region are actively investing in AI and has seen growth in AI-influenced revenue over the past years. From 2019 to 2022, the AI-influenced revenue of surveyed companies doubled, increasing from 8% to 19%. There is a strong prediction that this trend will continue, with AI-influenced revenue expected to increase significantly, reaching 26% by 2025. However, Nordic companies have yet to take full advantage of what AI can do for their business. 81% of surveyed organizations in 2022 were still experimenting with AI. Only 6% were considered AI achievers, meaning they have differentiated AI strategies and ability to scale value from AI.
There is less available data on the AI maturity of Icelandic companies, but looking at closely related data points can provide some indication. In international comparison, Iceland ranks 20th on the Global Innovation Index 2023 ranking. This means that Iceland outranks many large European countries such as Spain, Portugal, and Italy. However, both Denmark, Finland, Sweden, and Norway rank higher.
 Additionally, when looking at the Tortoise Media Global AI Index 2023, which benchmarks nations on their level of investment, innovation, and implementation of AI, Iceland is ranked the lowest among the Nordic countries at number 40. There is a gap of 16 rankings to the next Nordic country, Norway.
Tortoise Media, The Global AI Index, 2023 https://www.tortoisemedia.com/intelligence/global-ai/
 Based on this, it is assumed that Icelandic companies on average have less mature AI capabilities than companies from Denmark, Finland, Sweden, or Norway.

Adoption of Ethical and Responsible AI in the Nordics

Businesses in the Nordics are taking important steps to develop an ethical and responsible approach to AI, but few have operationalized it. In a study from 2023, 28% of surveyed business leaders of major enterprises in Denmark, Finland, Sweden, and Norway had developed formal ethical guidelines, but only 22% had operationalized them across their organization. 49% stated that they had not taken any action at all to develop ethical guidelines for AI.
Accenture, Re:humanize Institute, Impact leadership in the age of Generative AI, 2024 https://rehumanizeinstitute.org/wp-content/uploads/2024/01/Nordic-Responsible-Business-Report.pdf
Looking ahead, the maturity of AI adoption in the Nordics is expected to increase
, and during 2023, businesses started to pilot generative AI initiatives as well. In a survey from October 2023, 22% of surveyed businesses in Denmark, Finland, Sweden, and Norway stated they were piloting generative AI, whereas 25% had hired experts to systematically use generative AI.
Accenture, Re:humanize Institute, Impact leadership in the age of Generative AI, 2024 https://rehumanizeinstitute.org/wp-content/uploads/2024/01/Nordic-Responsible-Business-Report.pdf
 The growing focus on generative AI is likely to increase the focus on building capabilities to develop and deploy AI solutions ethically and responsibly. 68% of surveyed leaders from businesses in Denmark, Finland, Sweden, and Norway stated that the availability and development of generative AI has increased their interest in ethical and responsible AI
Accenture, Re:humanize Institute, Impact leadership in the age of Generative AI, 2024 https://rehumanizeinstitute.org/wp-content/uploads/2024/01/Nordic-Responsible-Business-Report.pdf
. Given this trend, if Icelandic companies have less mature AI capabilities, it can also be assumed that they are less mature in ethical and responsible AI than companies in Denmark, Finland, Sweden, and Norway.
Chapter3-1.png

Nordic Businesses’ Preparedness for Emerging AI Regulation and Ethical Requirements

There is an increase in new C-suite roles with a focus on AI strategy such as Chief Analytics Officer, Chief Data Officer, and Chief Digital Officer, that are working closely with the CEO and board members to among all prepare to meet emerging legal requirements
. In addition, Nordic businesses see attracting and upskilling AI talent as an important factor for operationalizing ethical and responsible AI and ensuring compliance with relevant regulations. They also recognize that they need to consider the full AI value chain and work with third-party AI model providers to fully mitigate AI risks.
Accenture, From AI compliance to Competitive Advantage, 2022 (Data extracted from respondents from Denmark, Finland, Sweden, Norway) https://www.accenture.com/content/dam/accenture/final/a-com-migration/r3-3/pdf/pdf-179/accenture-responsible-by-design-report.pdf
What Nordic organizations consider as the top three greatest internal challenges in meeting future regulations of AI:

1.

Lack of talent and expertise

2.

Lack of appropriate technical approaches

3.

Lack of clearly defined roles and responsibilities
Accenture, From AI compliance to Competitive Advantage, 2022 (Data extracted from respondents from Denmark, Finland, Sweden, Norway) https://www.accenture.com/content/dam/accenture/final/a-com-migration/r3-3/pdf/pdf-179/accenture-responsible-by-design-report.pdf
For generative AI, Nordic businesses are likely to build downstream applications based on foundation models, which means that they may need to consider their AI providers as part of their ethical and responsible AI approach. Simultaneously, developers of foundation models struggle with living up to transparency requirements.
In October 2023, no major foundation model developer was close to providing adequate transparency. Across 100 transparency indicators, the top-scoring model scored only 54
Stanford University’s Centre for Research on Foundation Models (CRFM), Institute on Human-Centred Artificial Intelligence (HAI), MIT Media Lab, Princeton University’s Centre for Information Technology Policy, The Foundation Model Transparency Index, 2023 https://crfm.stanford.edu/fmti/
. For Nordic businesses building generative AI solutions and use-cases, the current lack of transparency into foundation models could potentially hinder their ability to prepare for legal and ethical requirements posed on them by various stakeholders.

Ecosystem Collaborations for Ethical and Responsible AI

When asked about how to increase adoption of ethical and responsible AI among businesses in the Nordics, interviewed companies for this report stress among all the importance of complementing regulations with standards, guidelines, and methodologies. Sharing successful examples of how others have gone about to operationalize ethics and responsibility in AI solutions should be encouraged.
There are several national ecosystem initiatives in the Nordics focused on increasing collaboration on AI, many of these focus on technology development. For example, Iceland has created Almannarómur, an Icelandic language technology center tasked with protecting the Icelandic language in the digital world and providing access to language technology. Almannarómur is funded by the Icelandic government, and founding members include academic institutions and industry actors.
Almannarómur, About Almannarómur, 2023 https://almannaromur.is/en/um-okkur
 The Icelandic government is also collaborating with OpenAI, the company behind ChatGPT and the developer of the GPT-models, to use the GPT-4 model in preservation efforts of the Icelandic language
. There are more ongoing initiatives to offer large language models (LLMs) for the different Nordic languages. In Sweden, AI Sweden and the Swedish AI ecosystem has collaborated on creating a LLM for the Swedish language.
AI Sweden, GPT-SW3, 2023  https://www.ai.se/en/project/gpt-sw3
 Linköping University, a public research university in Sweden, is also working on developing a trustworthy open LLM for Nordic and Germanic languages
Linköpings Universitet, De tar fram en tillförlitlig ChatGPT för europeiska språk, 2023 https://liu.se/nyhet/trustllm
. The Finnish private AI lab Silo AI launched a consortium together with TurkuNLP, aimed at developing a family of open LLMs, including the world’s largest open source LLM
Silo.AI, SiloGen launches a consortium to build the world’s largest open LLM, 2023 https://www.silo.ai/blog/silo-ai-launches-a-consortium-to-build-the-worlds-largest-open-llm
. In November 2023, the consortium released the first model checkpoint for the model named Poro 34B, which is a 34 billion parameter LLM for English, Finnish and coding languages
Silo.AI, Poro – a family of open models that bring European languages to the frontier, 2023 https://www.silo.ai/blog/poro-a-family-of-open-models-that-bring-european-languages-to-the-frontier
. In Norway, the start-up Bineric AI has developed a LLM for Norwegian, NorskGPT
Bineric, Advancing Large AI Models, 2024 (see NorskGPT) https://bineric.com/
. Note that these are only selected examples as of January 2024. Given the pace of technology development, this list should not be considered exhaustive.
However, as identified in the Nordic AI and Data Ecosystem report from 2022 by Nordic Innovation, there is a lack of cross-Nordic ecosystem collaboration focused on helping industry actors understand how to adopt AI ethically and responsibly. It was recommended that the Nordics should focus on creating frameworks, guidelines, and networks to share best practices, use-cases, and knowledge between each other.
Nordic Innovation, The Nordic AI and Data Ecosystem, 2022 https://norden.diva-portal.org/smash/get/diva2:1667628/FULLTEXT02.pdf
 Nordic Innovation later launched the project that this report is part of, as an initiative to start mobilizing the Nordic AI ecosystem around the topic of ethical and responsible AI.
Nordic Innovation, The Nordics mobilize to advance adoption of ethical AI by building a supportive ecosystem, 2023 https://www.nordicinnovation.org/news/press-release-nordics-mobilize-advance-adoption-ethical-ai-building-supportive-ecosystem
 Looking at global examples, mobilizing ecosystem actors to collaborate on defining guidelines and methodologies has been important enablers of increased adoption of ethical and responsible AI.
One global example of broad ecosystem collaboration is the Monetary Authority of Singapore (MAS), that established the Fairness, Ethics, Accountability and Transparency (FEAT) principles in 2018 together with members from the financial sector. These were introduced to accelerate the adoption of responsible AI to enable public trust in AI used in financial institutions.
Monetary Authority of Singapore, MAS introduces new FEAT Principles to promote responsible use of AI and data analytics, 2018 https://www.mas.gov.sg/news/media-releases/2018/mas-introduces-new-feat-principles-to-promote-responsible-use-of-ai-and-data-analytics
 Subsequently, MAS initiated the Veritas Consortium in 2019, comprising of 27 industry actors. The Veritas Consortium has since developed several assessment methodologies and tools for implementing the FEAT principles. For example assessment methodologies have been developed for each of the FEAT principles - fairness,
, ethics and accountability
, and transparency
, as well as an open-source software
Monetary Authority of Singapore, MAS-led Industry Consortium Publishes Assessment Methodologies for Responsible Use of AI by Financial Institutions, 2022 https://www.mas.gov.sg/news/media-releases/2022/mas-led-industry-consortium-publishes-assessment-methodologies-for-responsible-use-of-ai-by-financial-institutions
. These methodologies empower businesses to apply the principles in their AI operations, define their own targets, identify specific attributes, and provide quantifiable measurements. This example highlights the benefits of broad industry collaboration on common challenges and is one example where industry collaboration for AI standardization has proven successful.
As Nordic companies mature their AI capabilities, the emphasis on ensuring ethical and responsible use of the technology is expected to intensify. Enabling collaboration and capability transfer can help make sure that companies leverage learnings from others, both within and outside of their industry.
chapter3-2.png
Nordic Council of Ministers I www.norden.org/publications I pub@norden.org