## 1. AI Preparedness ## Source details **Canonical URL:** [1. AI Preparedness](https://www.imf.org/-/media/files/publications/selected-issues-papers/2026/english/sipea2026077.pdf) ## Other formats - [Markdown version](/-/media/files/publications/selected-issues-papers/2026/english/sipea2026077.pdf.md) - [Structured JSON version](/-/media/files/publications/selected-issues-papers/2026/english/sipea2026077.pdf.json) --- ### A. Introduction and policy context - AI offers substantial long-term benefits (raising productivity, relieving workers from repetitive tasks, stimulating innovation, improving everyday convenience) but can create uneven short-term outcomes across society; some groups (for example, senior people with limited digital knowledge, or workers whose jobs are directly substituted by AI) may be worse off. - The Roadmap for Sweden produced by the AI Commission finds Sweden well positioned to benefit from AI but with gaps to close: - Strengths: fossil-free energy generation comparative advantage; cold climate attractive for hosting data centers; relatively skilled and adaptable labor force; one of Europe’s deepest venture capital markets; extensive high-quality long time-series of data. - Gaps: non-uniform implementation of EU regulations across member states; insufficient reskilling and retraining opportunities; persistent mismatch between private sector skill demand and educational outcome; obstacles to attracting and retaining international talent. - Sweden’s standing in global benchmarks: - Global AI Index from Tortoise Media: overall ranking for Sweden fell from 17th in 2023 to 25th in 2024, out of 83 countries included. - Sweden ranks particularly low at 57th in the area of government strategy (the recent published AI Strategy may elevate Sweden’s position). - National strategies: - Digitalization Strategy 2025–2030 structured around five strategic domains: digital competence, business digitalization, public administration, welfare services, and connectivity; cross-cutting areas include AI, access to data, and security. - A dedicated AI Strategy and accompanying action plan organized around three pillars: - AI for societal benefit: streamline legal and regulatory frameworks, expand data access and use, advance standardization, enhance security and defense, lead AI adoption in the public sector. - Sustainable development: promote human-centric AI, develop national language models, support climate and energy objectives, facilitate labor market adjustment and skills development. - AI for competitiveness and innovation: strengthen the AI ecosystem, foster business and entrepreneurship, scale up digital infrastructure and computing capacity, deepen research and expertise. ### B. Current uptake and enterprise usage - Among enterprises with more than 10 employees: - Share of AI usage rose from 10 percent in 2023 to about 35 percent in 2025. - Sweden’s adoption is similar to other Nordic countries and well above the EU average of 15 percent. - Adoption patterns: - Large companies lead adoption due to business needs and fixed costs associated with adopting AI. - Highest uptake by NACE 1-digit: Information Communication Technology (ICT) and Professional Services. - Among enterprises using AI: nearly three-quarters deploy it for automating machine movement and close to 60 percent for image processing. ### C. Labor market implications and measurement approach - Methodology: - Uses C-AIOE (Complementary-AIOE) index building on Felten, Raj and Seamans (2021) and Pizzinelli and others (2023) to measure occupational exposure to AI and whether AI complements or substitutes labor. - Applied to the 2023 EU Labor Force Survey (EU-LFS) at the 3-digit occupational level. - Classification: occupations with AIOE above the 3-digit median = high-exposure; complementary scores above the 3-digit median = high complementarity. Three mutually exclusive groups: HEHC, HELC, and LE. - Caveats: - Considerable uncertainty around impacts; empirical evidence remains tentative and task content may evolve as AI technologies develop. ### D. Key findings on exposure, complementarity, and vulnerabilities - Overall exposure and complementarity: - About 70 percent of the Swedish labor force falls into high-exposure occupations. - Among these high-exposure workers, roughly 70 percent are in occupations where AI is largely complementary. - Sweden and its Nordic peers outrank other EU countries on both exposure and complementarity measures. - HEHC occupations mostly consist of professionals across business, education, science and engineering, and sales. - ICT professionals: - All ICT professionals in Sweden belong to HELC occupations and face greater displacement risk. - Relevant 3-digit ISCO-08 occupations: software and applications developers (251) and database and network professionals (252). - These two occupations rank in the 26th and 29th percentiles, respectively, in complementarity (well below the median across 127 three-digit occupations). - Recent Swedish evidence: slower employment growth among young workers in AI-exposed occupations (Lodefalk and others, 2026); in ICT industry, nearly one-third of ICT graduates now have to work outside their field of training. - Skill imbalance and education pipeline: - Sweden exhibits one of the highest degrees of excess demand in the Skill Imbalance index among 36 countries, second only to Luxembourg. - Skill Readiness index: Sweden ranks 11th out of 24. - Skill Readiness components indicate Sweden’s weak capacity to supply graduates with new skills (component (1)), while maintaining leading positions in adult job-related training (component (2)) and current workforce proficiency (component (3)). - 2023/24 academic year enrollments: 75 percent of registered students enrolled in humanity and social sciences programs, while about 40 percent in science and engineering fields; students can enroll in multiple majors and about 40 percent of students choose double major. - Public versus private sector: - Public sector proxy (NACE O, P, Q—public administration, education, health care) employs about one-third of the workforce. - Close to two-thirds of public sector workers could experience productivity gains from AI adoption. - Less than 40 percent of workers in the private sector could experience productivity gains. - Gender breakdown: - Women: nearly 70 percent work high-exposure occupations, and more than two-thirds of these jobs exhibit strong complementarity. - Men: 62 percent work high-exposure occupations, and 65 percent of those jobs exhibit strong complementarity. - Vulnerable female groups include financial professionals, administrative secretaries, government regulatory professionals, and general clerks. - Vulnerable male groups are predominantly ICT professionals. - About 13 percent of female personal care and personal services workers, and 17 percent of male construction laborers, drivers, and machine operators are largely shielded from AI-related disruptions. - Socio-demographic vulnerabilities: - Less educated, low-income, young and immigrant workers are disproportionately unlikely to benefit from AI (jobs have low exposure or low complementarity). - These groups often overlap; in Sweden fewer than 10 percent of workers have less than lower secondary education. - Limited financial buffers among these groups make adjustment harder without targeted assistance. ### E. Policy implications and priorities - Facilitate labor mobility across regions and sectors. - Better align the education system with labor market needs to address the weak capacity to supply graduates with new skills. - Expand reskilling and retraining opportunities, especially for less-educated, low-income, young, and immigrant workers. - Address obstacles to attracting and retaining international talent. - Leverage public sector potential by leading AI adoption in public administration, education, and health care to realize productivity gains. - Pursue a holistic strategy backed by strong political leadership to close gaps in AI preparedness and implementation. --- ### 15. Sweden should continue to ### Overview - Nearly half of its labor force is well-positioned to receive a productivity boost from AI. - Swedish enterprises, on average, face fewer obstacles to deploying AI than their European peers. - Horizontal policies that promote technology diffusion and improve resource allocation (IMF, 2025; Productivity Commission, 2024, 2025) are equally relevant in supporting effective AI uptake. ### Main constraints to AI adoption - Difficulties in recruiting workers with the right skills. - Limited access to data. - Legal uncertainties, including around privacy protection. - Given that Sweden’s abundance of data is one of its key strengths in the AI ecosystem, clearer and more harmonized regulation, particularly the alignment with EU frameworks, is essential to easing the latter two barriers. - Implementing relevant measures outlined in the Roadmap, Digitalization Strategy, and AI Strategy would substantially advance this agenda. ### Education and skills policy recommendations - Adapting the education system offers a fundamental solution to addressing skill imbalances in the labor market. - Expanding the share of students choosing science or technology fields provides the most durable solution to evolving skill needs driven by technological changes. - Key measures include: - Raising the quality of pre-college education. - Improving guidance and information provided to students on choosing their fields of study. - Strengthening linkages between universities and industry. - Closer university-industry collaboration could: - Enhance the employment and earning prospects of graduates in science and technology programs. - Help alleviate the elevated unemployment pressures faced by labor market entrants. ### Active labor market policies and social protection - Active labor market policies with targeted support for highly vulnerable groups can smooth their adjustment cost. - Sweden already has strong capacity in upskilling and reskilling the workforce, but this can be further enhanced by expanding opportunities in: - Adult education. - Vocational education and training. - On-the-job learning closely aligned with firms’ skill needs. - The social safety net should strike an appropriate balance between cushioning transition costs and maintaining incentives for retraining (Ljungqvist and Sargent, 1998). - Labor market programs need to be better tailored to individuals outside the labor force, particularly those with low educational attainment and immigrants. ### Key citations and supporting literature (as referenced) - IMF, 2025; Productivity Commission, 2024, 2025. - Ljungqvist and Sargent, 1998. *Source: IMF staff summary of "1. AI Preparedness" (Selected Issues Paper, Sweden).* ### 1. AI Preparedness ________________________________________________________________________ 3 ### 1. AI Preparedness ### A. Introduction and policy context - AI offers substantial long-term benefits (raising productivity, relieving workers from repetitive tasks, stimulating innovation, improving everyday convenience) but can create uneven short-term outcomes across society; some groups (for example, senior people with limited digital knowledge, or workers whose jobs are directly substituted by AI) may be worse off. - The Roadmap for Sweden produced by the AI Commission finds Sweden well positioned to benefit from AI but with gaps to close: - Strengths cited: fossil-free energy generation comparative advantage, cold climate attractive for hosting data centers, relatively skilled and adaptable labor force, one of Europe’s deepest venture capital markets, extensive high-quality long time-series of data. - Gaps cited: non-uniform implementation of EU regulations across member states, insufficient reskilling and retraining opportunities, persistent mismatch between private sector skill demand and educational outcome, obstacles to attracting and retaining international talent. - Sweden’s standing in global benchmarks (example cited): - Global AI Index from Tortoise Media: overall ranking for Sweden fell from 17th in 2023 to 25th in 2024, out of 83 countries included. - Sweden ranks particularly low at 57th in the area of government strategy (the recent published AI Strategy may elevate Sweden’s position). - National strategies: - Digitalization Strategy 2025–2030 structured around five strategic domains: digital competence, business digitalization, public administration, welfare services, and connectivity; cross-cutting areas include AI, access to data, and security. - A dedicated AI Strategy and accompanying action plan organized around three pillars: - AI for societal benefit: streamline legal and regulatory frameworks, expand data access and use, advance standardization, enhance security and defense, lead AI adoption in the public sector. - Sustainable development: promote human-centric AI, develop national language models, support climate and energy objectives, facilitate labor market adjustment and skills development. - AI for competitiveness and innovation: strengthen the AI ecosystem, foster business and entrepreneurship, scale up digital infrastructure and computing capacity, deepen research and expertise. ### B. Current uptake and enterprise usage - Among enterprises with more than 10 employees: - Share of AI usage rose from 10 percent in 2023 to about 35 percent in 2025. - Sweden’s adoption is similar to other Nordic countries and well above the EU average of 15 percent. - Adoption patterns: - Large companies lead adoption due to business needs and fixed costs associated with adopting AI. - Highest uptake by NACE 1-digit: Information Communication Technology (ICT) and Professional Services. - Among enterprises using AI: nearly three-quarters deploy it for automating machine movement and close to 60 percent for image processing. ### C. Labor market implications and measurement approach - Methodology: - Uses C-AIOE (Complementary-AIOE) index building on Felten, Raj and Seamans (2021) and Pizzinelli and others (2023) to measure occupational exposure to AI and whether AI complements or substitutes labor. - Applied to the 2023 EU Labor Force Survey (EU-LFS) at the 3-digit occupational level. - Classification: occupations with AIOE above the 3-digit median = high-exposure; complementary scores above the 3-digit median = high complementarity. Three mutually exclusive groups: HEHC, HELC, and LE. - Caveats: - Considerable uncertainty around impacts; empirical evidence remains tentative and task content may evolve as AI technologies develop. ### D. Key findings on exposure, complementarity, and vulnerabilities - Overall exposure and complementarity: - About 70 percent of the Swedish labor force falls into high-exposure occupations. - Among these high-exposure workers, roughly 70 percent are in occupations where AI is largely complementary. - Sweden and its Nordic peers outrank other EU countries on both exposure and complementarity measures. - HEHC occupations mostly consist of professionals across business, education, science and engineering, and sales. - ICT professionals: - All ICT professionals in Sweden belong to HELC occupations and face greater displacement risk. - Relevant 3-digit ISCO-08 occupations: software and applications developers (251) and database and network professionals (252). - These two occupations rank in the 26th and 29th percentiles, respectively, in complementarity (well below the median across 127 three-digit occupations). - Recent Swedish evidence: slower employment growth among young workers in AI-exposed occupations (Lodefalk and others, 2026); in ICT industry, nearly one-third of ICT graduates now have to work outside their field of training. - Skill imbalance and education pipeline: - Sweden exhibits one of the highest degrees of excess demand in the Skill Imbalance index among 36 countries, second only to Luxembourg. - Skill Readiness index: Sweden ranks 11th out of 24. - Skill Readiness components indicate Sweden’s weak capacity to supply graduates with new skills (component (1)), while maintaining leading positions in adult job-related training (component (2)) and current workforce proficiency (component (3)). - 2023/24 academic year enrollments: 75 percent of registered students enrolled in humanity and social sciences programs, while about 40 percent in science and engineering fields; students can enroll in multiple majors and about 40 percent of students choose double major. - Public versus private sector: - Public sector proxy (NACE O, P, Q—public administration, education, health care) employs about one-third of the workforce. - Close to two-thirds of public sector workers could experience productivity gains from AI adoption. - Less than 40 percent of workers in the private sector could experience productivity gains. - Gender breakdown: - Women: nearly 70 percent work high-exposure occupations, and more than two-thirds of these jobs exhibit strong complementarity. - Men: 62 percent work high-exposure occupations, and 65 percent of those jobs exhibit strong complementarity. - Vulnerable female groups include financial professionals, administrative secretaries, government regulatory professionals, and general clerks. - Vulnerable male groups are predominantly ICT professionals. - About 13 percent of female personal care and personal services workers, and 17 percent of male construction laborers, drivers, and machine operators are largely shielded from AI-related disruptions. - Socio-demographic vulnerabilities: - Less educated, low-income, young and immigrant workers are disproportionately unlikely to benefit from AI (jobs have low exposure or low complementarity). - These groups often overlap; in Sweden fewer than 10 percent of workers have less than lower secondary education. - Limited financial buffers among these groups make adjustment harder without targeted assistance. ### E. Policy implications and priorities (as reflected in the source) - Facilitate labor mobility across regions and sectors. - Better align the education system with labor market needs to address the weak capacity to supply graduates with new skills. - Expand reskilling and retraining opportunities, especially for less-educated, low-income, young, and immigrant workers. - Address obstacles to attracting and retaining international talent. - Leverage public sector potential by leading AI adoption in public administration, education, and health care to realize productivity gains. - Pursue a holistic strategy backed by strong political leadership to close gaps in AI preparedness and implementation. *Source: IMF staff summary of "1. AI Preparedness" (Selected Issues Paper, Sweden).* ### 15. Sweden should continue to ### 15. Sweden should continue to ### Overview - Nearly half of its labor force is well-positioned to receive a productivity boost from AI. - Swedish enterprises, on average, face fewer obstacles to deploying AI than their European peers. - Horizontal policies that promote technology diffusion and improve resource allocation (IMF, 2025; Productivity Commission, 2024, 2025) are equally relevant in supporting effective AI uptake. ### Main constraints to AI adoption - Difficulties in recruiting workers with the right skills. - Limited access to data. - Legal uncertainties, including around privacy protection. - Given that Sweden’s abundance of data is one of its key strengths in the AI ecosystem, clearer and more harmonized regulation, particularly the alignment with EU frameworks, is essential to easing the latter two barriers. - Implementing relevant measures outlined in the Roadmap, Digitalization Strategy, and AI Strategy would substantially advance this agenda. ### Education and skills policy recommendations - Adapting the education system offers a fundamental solution to addressing skill imbalances in the labor market. - Expanding the share of students choosing science or technology fields provides the most durable solution to evolving skill needs driven by technological changes. - Key measures include: - Raising the quality of pre-college education. - Improving guidance and information provided to students on choosing their fields of study. - Strengthening linkages between universities and industry. - Closer university-industry collaboration could: - Enhance the employment and earning prospects of graduates in science and technology programs. - Help alleviate the elevated unemployment pressures faced by labor market entrants. ### Active labor market policies and social protection - Active labor market policies with targeted support for highly vulnerable groups can smooth their adjustment cost. - Sweden already has strong capacity in upskilling and reskilling the workforce, but this can be further enhanced by expanding opportunities in: - Adult education. - Vocational education and training. - On-the-job learning closely aligned with firms’ skill needs. - The social safety net should strike an appropriate balance between cushioning transition costs and maintaining incentives for retraining (Ljungqvist and Sargent, 1998). - Labor market programs need to be better tailored to individuals outside the labor force, particularly those with low educational attainment and immigrants. ### Key citations and supporting literature (as referenced) - IMF, 2025; Productivity Commission, 2024, 2025. - Ljungqvist and Sargent, 1998. *SWEDEN INTERNATIONAL MONETARY FUND* --- _Source: https://www.imf.org/-/media/files/publications/selected-issues-papers/2026/english/sipea2026077.pdf_