Swedish report: Strengthened competitiveness through the use, development and ecosystem of AI in Sweden

TechSverige has launched a report entitled “With AI as a means – Strengthening competitiveness with the use, development and ecosystem of AI in Sweden”. In March 2024, TechSweden held a series of roundtable discussions to discuss AI development, use and ecosystems in the country. AI progress in Sweden is at a crucial stage with more AI solutions available on the market, established EU rules and ongoing government initiatives, including a new AI Commission and a digitalisation strategy. Initiatives such as AI Sweden and WASP play an important role, but further action is needed. The aim of the report is to present concrete proposals to strengthen Sweden’s competitiveness through AI development and use.

Table of contents

Foreword

• A crucial stage for AI in Sweden

The tech industry and AI development

• The tech industry is central to AI development in Sweden
• AI

Role in the development of society

• The need to strengthen Sweden’s international position
• The EU does not offer the whole solution
• Demand for AI skills exacerbates skills shortages
• AI means broader changes, not just technological ones

Strategic proposals for Sweden

1. Improved data availability and use of synthetic data
2. Ensuring computational power for AI
3. Reduce compliance costs and resources
4. Strengthen investment and promote the use of AI in business
5. Government efficiency and quality improvement through AI
6. Ensuring AI competence in Sweden
7. Creation of 10 high-profile innovation awards for AI
8. Promoting international cooperation and sustainability

Summary

• A summary of the proposals and the way forward

The tech industry is crucial for AI development in Sweden

Despite economic challenges, the importance of tech companies for the Swedish economy is growing. In 2022, the tech sector contributed almost SEK 350 billion to GDP, which is more than the traditional basic industry. The tech industry accounts for 11 percent of Sweden’s exports, making it one of the most important export segments. Between 2020 and 2023, the tech sector created 33 000 new jobs and employed over 265 000 people in the second quarter of 2023.

The sector plays a central role in people’s everyday lives, welfare, the climate transition and for productivity and innovation in other industries and the public sector. However, this positive development is not self-evident as global tech competition increases.

TechSverige’s report “Swedish tech industry 2023” shows that the growth in tech in countries such as Denmark, Finland, the Netherlands and Austria has been twice as fast as in Sweden between 2015 and 2020. Several of these countries are also ahead of Sweden in AI.

Sweden’s lead is decreasing due to the lack of reforms and stagnating digitalisation in the public sector. Europe is also lagging behind the US and Asia, especially in AI where development and competition are increasing.

In order for companies to take advantage of new technology and translate it into innovations and increased productivity, basic conditions are needed. Reforms alone do not guarantee success, but companies need space to experiment and evolve to turn AI power into competitiveness.

AI and societal development

Discussions on AI have intensified, especially with the recent impact of generative AI and large language models. However, it is important to remember that AI is an umbrella term that is already widely used, for example in healthcare for image analysis and in the energy sector for optimization.

TechSweden, as an employer organisation, is involved in the transition issues related to AI. Historically, Sweden has managed technological change by rapidly adopting new technologies, implementing reforms and investing in skills development.

We are now facing a new transition, similar to that when the internal combustion engine and the internet were established. These changes have previously strengthened Sweden’s innovation capacity, export industry and ability to participate in global trade. With AI as a tool, Sweden can once again strengthen its competitiveness and become a key player in global AI development.

Sweden’s international position needs to be strengthened

Sweden has the potential to become a leading player in AI, with strengths such as access to high-quality data and the ability to create AI applications. However, competing in AI research and development requires significant resources. It is important for Sweden to have ambitions in this area and to cooperate both within the EU and internationally.

Many of Sweden’s global companies need AI to maintain their competitiveness. The tech industry plays a key role here, but there are challenges. A survey by EY shows that only 25 percent of Swedish companies invest in generative AI, compared to the global average of 43 percent. In addition, only 36% invest in “analysis and AI”, below the global average of 63%.

Despite these challenges, a Demoskop survey from the end of 2023 shows that Swedish business leaders are optimistic about AI development. Most people believe that AI will increase efficiency and improve competitiveness.

However, Sweden ranks 17th in the Global AI Index, after countries such as Switzerland, Finland, the Netherlands and Denmark. The best in Europe is the UK, which ranks fourth globally, after the US, China and Singapore.

The EU does not have the full answer

The EU has implemented the AI Act to regulate AI, but global developments are moving faster than the EU’s efforts. Countries such as Canada, China and the United States are investing heavily in AI. Canada invests SEK 16 billion in AI infrastructure, while China accounts for 61 per cent of all AI patents globally. The US is a leader in AI with $67 billion in private sector investment in 2023, compared to $8 billion in China and $11 billion in the EU and the UK. These developments show that AI is rapidly becoming crucial for economic growth.

Skills shortages and AI

Sweden is already facing a skills shortage that makes it difficult to take full advantage of the benefits of digitalisation, and the increased demand for AI skills is exacerbating the situation. The areas that are growing fastest in the tech industry are AI Science and Data Science, and according to TechSweden’s skills survey (2024), the need in these areas is expected to increase by over 15 percent per year over the next 3 to 5 years. This demand is mainly for skilled workers. In addition to technical expertise, a broader understanding of AI is also needed to be able to implement the technology effectively in different businesses.

AI is not just technology – broader change is needed

In the long term, technological development contributes significantly to growth and prosperity. However, in the shorter term, technological advances can create social and economic challenges, such as threshold problems and conflicting objectives. Making full use of new technologies often requires complementary investments and innovations in companies and organisations, as well as new knowledge, ways of working and processes. Companies and organizations must constantly experiment to find optimal working methods and business models. The demand for new technological opportunities is also changing, leading to major structural changes in society.

Policy initiatives should both lower the thresholds for experimenting with and introducing new technologies and address the obstacles and conflicting objectives of structural change. AI can play an important role in meeting the major challenges of our time, such as tackling environmental and climate issues and strengthening and future-proofing welfare.

The policy objective should not be technology per se, but how technology can contribute to solving current challenges and achieving societal goals. TechSweden presents proposals in eight areas to promote the use, development and ecosystem of AI in Sweden. AI is a necessary tool to strengthen and renew Sweden’s competitiveness.

AI as a driving force – Strategies for a competitive Sweden

In order for Sweden to compete globally, it is crucial to identify areas where AI can create the greatest benefit for industry. The government should work in a structured manner with industry to identify sectors and projects where reforms can bring market benefits.

Sweden’s ability to use AI to increase GDP will contribute to a sustainable social transition, improve welfare, strengthen competitiveness and create new jobs. It is essential that the whole business community is enabled to experiment with AI, and that policies identify and remove barriers and lower thresholds.

1. Access to data and synthetic data strengthens AI capabilities

The availability of high-quality data, especially in the welfare sector, represents an important competitive advantage in the data-driven economy. However, Swedish datasets are spread across many authorities, regions and municipalities, which means that the thresholds for working with data from the public sector are high. Frictionless data sharing is required, and small businesses need special support with data management in smaller value chains.

Synthetic data, created artificially instead of being the result of real events, is becoming increasingly important. A data pool could serve as a central entry point for both the public and private sectors, working on increasing interoperability in the public sector and making high-quality data available to private actors.

All authorities, regions and municipalities should be encouraged to provide open APIs for data that can already be made available, allowing users to continuously download updated data in real time.

Proposal:

• A data pool can act as a trusted third party for data intermediation and management, reinforcing open data work.
• Establish live regulatory sandboxes that include multiple parties in a data-driven value chain.
• Explore a corresponding data pool at Nordic level to expand the dataset and strengthen resources.
• An investigation or cooperation should be established to make public sector data available, create synthetic data and promote their use. A strategy should be formulated and integrated into the national data pool.

2. Ensure access to computational power for AI

Developing and deploying AI requires several components, including skills, appropriate policy and legal frameworks, access to data, and sufficient computational power to manage machine learning models, advanced algorithms and large data sets. Many countries, including Sweden, lack complete knowledge of what capabilities exist and what is needed to meet their AI strategies and be competitive.

It is crucial that both industry and the public sector have good access to computational power in order not to limit the deployment of AI solutions on a large scale. Developments in this area are fast and computational power is increasingly offered as a service, which makes it important to distinguish between private and common computational power and resources. Lack of computational power must not become a bottleneck for innovation and development.

Proposal:

• In collaboration with industry, the government should investigate both available and necessary computational power based on various future scenarios and strategies.
• A special effort should be made to strengthen the infrastructure for computational power and High-performance Computing (HPC) already now, with private and public co-financing, as it is already clear that the infrastructure is too weak.

3. Reduce compliance costs and resources

An authority should be tasked with investigating and reducing compliance costs stemming from existing and new regulatory frameworks for digital markets, data and AI. The regulations are determined at EU level, but Sweden can, through implementation and complementary services, work to reduce the thresholds, especially for smaller companies. The National Board of Trade has carried out an impact assessment of several of these regulations, but concrete measures are now needed to facilitate digital transformation and the use of data and AI in Sweden.

Proposal:

• Further action is needed to lower legal and practical thresholds for data exchange and increased use of AI. The proposed authority may provide evidence or test contracts for data sharing and data flows between businesses and between businesses and public entities. The assignment can also be transferred to the soft infrastructure of the national data pool.
• Gather the responsibility. whereas the AI Regulation requires a dedicated AI authority, and this responsibility should also include the task of minimising compliance difficulties; This should not be split between several different actors but given to a specific authority.
• Launch experiments in some rule-intensive authorities to use AI to reduce the administrative burden on businesses. AI and digital tools should be used to facilitate redress and compliance. By training AI on rules and creating appropriate interfaces for data transfer, government processes can be streamlined and compliance costs reduced.
• Make regulatory interpretation in the public sector more consistent and predictable. One lesson learned from the introduction of the GDPR was that the interpretation of the rules was fragmented both within and between municipalities and regions. Smaller municipalities in particular do not have the resources to handle rule interpretation and compliance for both existing and future regulations regarding digital markets, data and AI. A central function for harmonised rule interpretation, advice and skills development around the new regulatory frameworks such as the AI Act is needed. This must, of course, be done while maintaining legal certainty, public authority responsibility and the right to appeal.

Source: Commercial College – The EU Single Market in the Digital Era

4. Strengthen investment and promote the use of AI in business

TechSverige conducted a survey in 2023 that showed that companies with medium or high AI knowledge, and who have already started the introduction of AI, experience mainly positive effects and plan more AI investments. On the other hand, companies with lower AI knowledge are more cautious. In order to accelerate the use of AI and lower the thresholds for companies, measures are needed to facilitate the creation of benefits from AI.

Promote innovative SMEs through tax incentives for the use of AI. Just like the importance of reducing the cost of compliance, the thresholds for using AI should be lowered, for example by making computational power available. One proposal is to introduce a tax deduction of 50 percent (capped) through a so-called GPU reform for the use of AI. This would benefit SMEs by lowering the thresholds for adopting AI and strengthening their innovation capacity. An inspirational model could be the tax reduction for the installation of green technology that has promoted the transition to climate-smart solutions.

promoting a development-driven economy and a competitive basic industry; In order to support the ability to innovate and develop AI applications, an enhanced R&D deduction with a focus on AI should be introduced.

* 8 GPUs (Graphics Processing Unit) are a type of processors that have proven very useful and efficient for resource-intensive AI applications.

5. Entrust authorities with streamlining and improving the quality of AI

AI, if used correctly, can provide significant efficiencies in areas such as customer service, administration, availability, service, and energy and heat consumption. To achieve this, both governance and resources are needed.

• Mandates to authorities: All relevant authorities should review their activities to identify opportunities for efficiencies using AI, for example by investigating which services and tasks can be enhanced and developed by AI. A higher level of ambition would be to establish a central function in the Government Offices to identify efficiency measures and coordinate these between authorities. This can foster synergies and exchange of experience.
• Financial support: A special pot of money should be set aside that authorities, in addition to their regular budget, can seek to develop and deploy innovative AI solutions that lead to higher efficiency, better results or other clear benefits for the business. Where appropriate, there should also be a requirement to disseminate and make available experiences and solutions.

6. Secure AI competence for Sweden

At a time when technological advances are continuously transforming working life and tasks, it is becoming increasingly important to have both specific domain knowledge and technical skills. To meet these needs, it is crucial that higher education institutions and universities of applied sciences integrate AI perspectives and applied AI into existing and new education in all areas.

• AI Education Lab: The government should set up a laboratory where researchers, teachers, principals and business actors work together to investigate and test how new technologies can improve the quality of primary and secondary education.
• Integrating AI in higher education: Universities and polytechnics need to integrate AI perspectives and applied AI into both existing and new education. The government should allocate specific resources to support higher education institutions and universities of applied sciences in this adaptation and development.
• Application-driven AI research: Introduce a dedicated programme for application-driven and research-oriented AI innovation. A gathering of forces similar to WASP, but focused on the application of AI, can make Sweden stronger in the use of AI.
• Skills development: To promote investment in employees’ digital skills, the Government should investigate a tax deduction for skills-enhancing activities in digitalisation, AI, information and cybersecurity. This would give employers stronger financial incentives to invest in the digital skills of their staff.
• International Talent Attraction: To attract cutting-edge skills and become a leader in AI, targeted policy efforts are needed. The government should adopt a strategy for tech talent, introduce six-year work permits and cap processing times for tech work permits at ten days.

7. Set up 10 high-profile innovation awards for AI

In order to exploit Sweden’s potential in the AI economy through access to high-quality datasets, the government can institute innovation awards for smart AI solutions. These prizes will target areas where public sector data can be used to solve complex problems that require significant innovation. Problems should be of such severity that they require technological leaps or breakthrough solutions, which can generate large gains in welfare, energy, climate and other key areas.

The advantage of such innovation prizes is that the competitors take on the risk, as the prize money is paid out only when the problem is solved. In addition, the awards can help build a vision of how AI can improve society.

• Proposal: The Government should announce ten innovation prizes worth SEK 11 million (equivalent to the Nobel Prize) to utilise public data investments and solve societal problems.

8. International cooperation and sustainability

Although the EU regulatory framework for AI is now in place, work continues in the EU and other international fora. In this context, the Government must safeguard Swedish interests, particularly in the green and digital transitions, where the potential of AI should be maximised. In Sweden, where further emission reductions are costly, this is particularly relevant.

Proposal:

• The Government should make it easier for Swedish actors to participate in international cooperation projects for AI development and standardisation. This includes collaborative research and the design of global regulatory frameworks that promote innovation and protect individual rights.
• A common Nordic platform should be developed to collect, anonymise and make available data from both the public and private sectors for AI research and development. The platform shall follow strict data protection and ethics guidelines that are accepted by all Nordic countries. Cooperation on harmonising rules and guidelines for AI and data sharing across Nordic borders will also be promoted to facilitate seamless data sharing and ensure high standards of privacy and data security. This can also be combined with the proposed national data pool.
• The government should further integrate AI into its climate work, for example by including it in the Climate Action Plan to better manage and mitigate climate change.
• Regions and municipalities should start using AI to make their operations more efficient. For example, AI can be used to optimise energy consumption in buildings, improve weather and climate forecasting, and streamline the use of renewable energy sources.

Using AI as a means Strengthened competitiveness with the use, development and ecosystem of AI in Sweden

Source:  European Digital Skills & Jobs Platform