Does the IMF AI Preparedness Index Tell You Where to Move for Work?

The short answer: the index is context, not a relocation instruction

No. The IMF AI Preparedness Index (AIPI) offers broad context about national conditions that may support AI adoption; it is not a relocation instruction for an individual. A stronger score may indicate more enabling capacity, but it does not establish that your occupation has more vacancies, that employers are adopting AI in ways that benefit your role, or that workers face less task change. The IMF’s “AI Preparedness Index (AIPI)” documentation calls the measure indicative and advises against using it for rankings. Its unit is the country, not a worker, city, employer, or job opening.

Use readiness as one contextual variable. Then ask whether your task bundle fits work employers are hiring for, whether the requirements are attainable, and whether a move works with your authorization, income needs, language, health, and family constraints. A country-level measure cannot answer those questions. A higher score cannot settle them, and a lower score cannot rule out a suitable role. The useful decision is not “Which country wins the index?” but “What evidence would make this move workable?” Do not relocate or reject a destination on AIPI alone; use it, at most, to frame questions after checking tasks, local openings, and feasibility.

Sources: AI Preparedness Index (AIPI), IMF

What does the IMF index measure—and leave out?

The IMF AI Preparedness Index (AIPI) is a country-level measure of conditions the IMF considers relevant to adopting AI across an economy. It combines indicators grouped under digital infrastructure, human capital and labor market policies, innovation and economic integration, and regulation and ethics. That makes it context about national capacity, not a guide to where one worker should move. The IMF's AI Preparedness Index (AIPI) documentation describes the measure as indicative, says it includes perception-based indicators, and advises against using it for ranking. A higher value is not a verdict that a country has more openings in your occupation, that your work is safer there, or that you should relocate. The construction explains what the measure can summarize and what it loses. The IMF normalizes available sub-indicators to a 0–1 scale using the full country sample, averages them within each of four dimensions, then takes a simple average of the dimensions for the overall index. These cover varied conditions: connectivity and online infrastructure; education and digital skills; research and innovation; and aspects of legal frameworks and governance. Some inputs are official or hard data, while others come from perception surveys or expert judgment, so the composite is not made of one uniform type of observation. The documentation also describes principal-component aggregation as a robustness check, reporting outcomes equivalent to simple averaging. This makes a broad composite possible. It does not turn the output into a count of AI jobs, a wage measure, or an estimate of how many employers in a city use AI. A single value compresses national conditions, so two places could reach similar values through different combinations of infrastructure, skills, innovation, and legal conditions. Even a dimension-level figure remains an aggregate. It cannot show whether capability is concentrated in your sector or city, or whether a particular employer has changed its workflow. Nor does it tell you whether your experience, credentials, language skills, and responsibilities fit the roles available. Those are the evidence that makes a location relevant to a specific worker. The IMF documentation states coverage of 174 countries and an annual index as of 2023. Inputs use the latest available year for each indicator, so not every component is established as measured in the same year. The document says economy-wide preparedness is difficult to measure because institutional requirements remain uncertain. A live IMF DataMapper address does not establish a newer observation period; the opened AIPI DataMapper dataset page did not render indicator details in its visible text. Cite the documented coverage and vintage unless downloadable data and metadata verify a later one. There is a reason to keep the index in view, provided it stays in its lane. National connectivity, human capital, innovation capacity, and legal frameworks can frame questions about conditions for AI adoption. But the index does not identify who captures productivity gains, whether employers are hiring, or how benefits and costs are distributed across workers and industries. Treating enabling conditions as proof of local opportunity would go beyond what the measure observes. For relocation, use AIPI as background only after a destination is plausible on more direct grounds. Identify the work you want to pursue and the tasks that make it valuable. Then check current openings and requirements, employer-specific evidence, authorization, language, credentials, compensation needs, housing, health, and family obligations. AIPI may frame questions about national conditions, or distinguish otherwise comparable options. It cannot rank your personal prospects. Its useful role is context; task fit, local hiring evidence, and constraints determine whether that context matters to your next move.

Sources: AI Preparedness Index (AIPI), IMF; AI Preparedness Index (AIPI), IMF DataMapper

Why readiness, exposure and hiring are different questions

Readiness asks whether national conditions may support AI adoption; exposure asks which tasks overlap with capabilities; adoption asks whether organizations use tools in real workflows; labor demand asks whether employers are hiring for work as it changes. These questions connect, but one answer cannot substitute for another. None tells an individual, by itself, whether moving is sensible. A national index can frame context, task research can identify where change may be relevant, and employer or vacancy evidence is closer to actual work. Personal feasibility remains a separate test.

Each step involves a conversion that may not happen. Enabling infrastructure does not show that a particular organization has chosen a tool or changed a workflow. Demonstrated capability does not establish reliable performance under a firm's data, security and accountability requirements. A pilot may not become routine use; implementation can stall or require checking. Use does not establish whether an employer will expand hiring, reduce hours or redesign roles.

The IMF staff discussion note, “Gen-AI: Artificial Intelligence and the Future of Work,” helps distinguish exposure from complementarity. Exposure concerns the potential relevance of AI capabilities to occupational work; complementarity concerns whether AI may support tasks while human input remains important. Both are lenses on what could change, not counts of jobs created or removed. Drafting a routine document, checking unusual cases, explaining a recommendation to a client and accountability for an error may relate to a tool in different ways. An occupation-level label compresses that variation. It cannot settle how an employer allocates tasks or what skills a worker will need after a redesign.

The International Labour Organization's 2025 “Generative AI and Jobs: A Refined Global Index of Occupational Exposure” shows why task research is a more informative middle rung than a country score, while remaining bounded. Its method draws on a representative sample of 29,753 tasks in Poland's occupational classification, a survey of 1,640 employed people across occupational groups, 52,558 task assessments, expert discussion and model predictions. It reports four exposure gradients; clerical work remains most exposed, and transformation is judged more likely than whole-job automation. These are modeled occupational findings using mixed human and model inputs. They do not observe a named employer's adoption, predict local vacancies, or assign an individual's displacement chance. The comparison of transformation and automation is the paper's assessment, not a personal probability.

A decision sequence should move from broad context toward evidence closer to the choice. AIPI belongs at the national-context level: it can prompt questions about enabling conditions, but cannot rank personal prospects. Occupational and task research can identify parts of a target role to examine. Employer workflow evidence can show whether particular uses are deployed; current vacancies can reveal advertised requirements in a specific market. Licensing, language and work authorization affect access. Salary, housing, health and family constraints test feasibility. Each layer answers a different question; none fills gaps in another.

This is an editorial reasoning framework, not a validated prediction model. It produces no combined score, and the evidence does not support weighting these factors into personal odds. An exposure measure can suggest task overlap while local postings still request domain expertise: both can be true because one concerns potential change and the other advertised requirements at a point in time. A high-readiness country with few suitable openings does not contradict the index; the index was not a hiring measure. Locate the unresolved question: what AI can do, whether employers use it, whether demand exists, or whether the worker meets requirements. Then gather evidence for that question instead of averaging unlike measures into a synthetic verdict.

Sources: Gen-AI: Artificial Intelligence and the Future of Work, IMF Staff Discussion Note SDN2024/001; Generative AI and Jobs: A Refined Global Index of Occupational Exposure, ILO Working Paper 140

What task exposure can tell you that a country score cannot

Task research can identify recurring activities that may be changed or assisted by generative AI, revealing differences hidden inside an occupation. It cannot establish that a named employer has adopted a tool, that a role will disappear, or that another country offers you a better job. National preparedness describes broad adoption conditions; occupational exposure estimates overlap between work activities and technological capabilities. Neither reports local vacancies or personal prospects.

The 2025 ILO Working Paper, “Generative AI and Jobs: A Refined Global Index of Occupational Exposure,” offers a more task-centered estimate. It uses a representative sample from the 29,753 tasks in Poland's occupational classification and surveys of 1,640 employed people across one-digit ISCO-08 groups. These yielded 52,558 assessments covering 2,861 tasks. Researchers compared this input with international expert surveys and Delphi-style discussions, then used the combined task knowledge to develop model predictions for ISCO-08 tasks. Its framework presents four increasing exposure gradients. This estimates task overlap; it does not count employers using AI or observed job losses.

The ILO summary says clerical occupations have the highest exposure, with increased exposure also appearing in some strongly digitized professional and technical work. It judges job transformation more likely than full automation because most occupations contain tasks requiring human input. That aggregate conclusion does not mean all workers in an occupation will have the same experience, or that human input guarantees demand. Exposure points to work worth examining; it cannot show what an employer will implement, how a role will change, or whether resulting work will be hired.

Describe work through deliverables and recurring tasks, not just a job title. Consider a monthly performance report: collecting figures, formatting a draft, investigating an unexpected movement, choosing a meaningful comparison, explaining uncertainty, and standing behind a recommendation are different activities. Some may be digitally repeatable; others depend on context, judgment, communication, or accountability. This is an illustration, not a measured case. A plausible draft is also different from an accurate result that can be checked at acceptable cost and fits workplace systems and rules.

The IMF Staff Discussion Note “Gen-AI: Artificial Intelligence and the Future of Work” likewise treats occupations as task bundles that may be replaced or complemented, with social, ethical, physical, and skill contexts shaping the relationship. It cautions that similar occupations may have different task mixes across countries and that firm adoption and reorganization are hard to foresee. An exposure estimate therefore cannot reveal how a target employer divides reporting, review, and decision responsibility, or whether local organizations have integrated AI into that workflow.

Use exposure to investigate a task, not as a migration trigger. Map one consequential work product: note repeatable digital steps, exceptions requiring investigation, decisions needing domain judgment, points of personal accountability, and checks before the output can be used. Then identify capabilities that could strengthen your contribution, such as verifying outputs, improving data quality, or explaining a recommendation. Test these against your role and local vacancies. High exposure signals where closer attention may help; evidence about actual roles, employers, requirements, and feasibility must guide whether to upgrade in place, explore an adjacent role, or consider moving.

Sources: Generative AI and Jobs: A Refined Global Index of Occupational Exposure, ILO Working Paper 140; Gen-AI: Artificial Intelligence and the Future of Work, IMF Staff Discussion Note SDN2024/001

Why adoption and labor demand can move in more than one direction

AI use can change what employers ask workers to do without producing one direction for employment. Some skills appear more often in vacancies for AI-exposed occupations, while selected skill demands decline in establishments with greater AI exposure. These findings can coexist: one describes advertised roles across occupations; the other compares hiring patterns within workplaces. Neither establishes whether a destination is hiring for your role, and neither can be read from a country's preparedness score. In *Artificial intelligence and the changing demand for skills in the labour market*, the OECD analyzes Lightcast online vacancies across ten OECD countries over a decade. In high-exposure occupations, the share of vacancies asking for at least one cognitive, emotional or digital skill rose by eight percentage points on average; social-skill demand rose by more than six points, and management and business-process skill demand by more than five. These are vacancy shares and trends in sampled markets, not net new jobs, proof that AI caused every change, or a forecast for an individual country. The report notes that its vacancy data did not directly observe AI use; occupational exposure served as a proxy. The report asks a different question with an establishment panel intended to better isolate differences associated with AI exposure. Greater exposure was associated with small declines in demand for several skill groups, including cognitive, digital, business-process and resource-management skills. The magnitudes were modest against rising demand for these skills overall. The analysis also found some evidence of increased demand for production, technology and physical skills. The authors suggest productivity effects and spillovers to other workers as possible explanations, not settled outcomes in every firm. An occupation-wide vacancy trend and a workplace comparison have different units; combining them into a single “AI creates” or “AI removes” verdict would erase what each measured. Several mechanisms could produce this mixed picture. If a tool takes over a repeatable step, an employer may need fewer hours of that task or ask existing workers to handle more cases. If lower costs or faster delivery let a business serve more customers, demand for complementary work could grow; that scale response is possible, not automatic. New checking, exception handling or coordination may also be needed, while implementation costs or weak integration can slow use. These are hypotheses for reasoning about change, not claims that a particular employer has adopted AI or plans to hire. Capability, actual use, task redesign and hiring remain separate questions. Worker reports add context about experience, not a labor-market forecast. In *Using AI in the workplace*, the OECD reports that four in five workers in its employer and worker surveys said AI improved performance, while three in five said it increased enjoyment. The same paper records concerns about work intensity, data use and inequality. Perceived benefits and risks can coexist; the reports do not prove causal productivity gains, net job growth or what workers in a specific city would experience. For a location decision, treat cross-country skill patterns as prompts for investigation. For that occupation, inspect local postings: which tasks, skills, seniority, language and credentials recur? Then seek employer-specific evidence of routine workflow use, rather than relying on broad AI familiarity in a listing. Postings are incomplete and stated requirements do not guarantee actual practice, but they are closer to local demand than a national readiness index. AIPI describes enabling context; it cannot tell which mechanism dominates in a target market or whether resulting work fits your constraints.

Sources: Artificial Intelligence and the Changing Demand for Skills in the Labour Market, OECD; Using AI in the Workplace, OECD

How should you compare two possible destinations?

Imagine a worker whose recurring task is preparing a weekly performance report: gathering figures, checking inconsistencies, explaining changes and answering colleagues’ questions. They are considering two locations where moving is potentially feasible. The useful comparison is not “Which country has the higher AI score?” It is “Where is this work being hired, what would employers expect, and can the worker realistically take those roles?” The IMF’s [AI Preparedness Index documentation](https://www.imf.org/external/datamapper/AIPINote.pdf) describes national conditions relevant to AI adoption. It can frame questions about infrastructure, skills, innovation and institutions, but it does not report vacancies for this occupation or resolve personal fit. Treating its value as a destination ranking asks the measure to answer a different question from the one it was built to describe. Start with the work, then examine evidence at the level where the decision will be made. Look at a manageable, recent sample of postings for the same role and seniority in each location. Record recurring requirements: domain knowledge, data quality checks, explaining results, tools, language, credentials and experience. A posting that says “AI skills” does not establish routine deployment; look for a concrete workflow, responsibilities and review expectations. Task research helps interpret requirements, but remains a separate rung. The ILO’s [Generative AI and Jobs: A Refined Global Index of Occupational Exposure](https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure) draws on task assessments, worker input, expert discussion and model predictions to estimate occupational exposure. It can identify activities that may be affected, but does not show that a specific employer has adopted a system or that a vacancy will change. The OECD’s [Artificial Intelligence and the Changing Demand for Skills in the Labour Market](https://www.oecd.org/content/dam/oecd/en/publications/reports/2024/04/artificial-intelligence-and-the-changing-demand-for-skills-in-the-labour-market_861a23ea/88684e36-en.pdf) examines online vacancies across ten countries and reports patterns at occupation and establishment levels. Those comparisons illuminate skill demand; they are not a current forecast for either destination or this worker. Apply hard constraints before weighing broad context. Can the worker obtain authorization to work? Are language expectations, credential recognition, salary floor, housing, healthcare and family needs workable? These are not footnotes to an AI score: if a role cannot be accessed or the move sustained, national readiness does not make it practical. Consider an example in which Location A appears stronger on an AIPI dimension, but relevant postings require credentials the worker lacks and offer few roles matching their reporting experience. Location B may have a less favorable national signal yet show more suitable openings, accessible requirements and workable costs. That would not prove B is a better labor market overall; it would make B more plausible for this worker’s decision, pending verification. Reverse the evidence and the conclusion could reverse too. AIPI may add context after both destinations pass these tests. If role demand, access and feasibility are otherwise comparable, differences in national infrastructure or institutions could prompt questions about training access or digital systems. But that is tie-break context, not a prediction that local hiring will be stronger or workers will capture productivity gains. The practical rule is simple: compare the same role bundle, inspect local evidence, clear personal feasibility gates, then use preparedness to enrich the picture. No composite score can decide which trade-offs are acceptable to the worker.

Sources: AI Preparedness Index (AIPI), IMF; Generative AI and Jobs: A Refined Global Index of Occupational Exposure, ILO Working Paper 140; Artificial Intelligence and the Changing Demand for Skills in the Labour Market, OECD

What should you verify before making a move?

Before spending money or uprooting routines, verify three things: the target work exists in a place you can access, you meet or can close its requirements, and the move fits your practical constraints. A national preparedness score cannot establish any of these. Use a small evidence check to decide whether to upgrade your current role, explore an adjacent one, or investigate relocation further. Start with one recurring work product, such as a monthly forecast, client brief or compliance review. Break it into steps: what is repetitive and digital, what requires judgment or context, and what carries accountability if it is wrong. Note where AI could assist and where a person must check or own the result. This map is not a prediction of what your employer will automate. Next, compare a bounded sample of current postings for the same target role in each location you could genuinely consider. Record seniority, required credentials, language, tools, domain knowledge and work outputs. Online vacancies are only a partial view of hiring, and a posting shows that an employer advertised a role, not that it remains open, that the employer uses AI in production, or that you would be selected. Vacancy and workplace evidence, such as that examined in the OECD report “How is AI changing the way workers perform their jobs and the skills they require?”, is closer to a work decision than a country composite, but does not forecast hiring chances in a particular city. Then check the gates a job listing cannot settle for you. Confirm work authorization through current official sources, and verify whether credentials or regulated qualifications transfer. Price housing and moving costs against a salary floor you can accept; include taxes and healthcare where relevant. Consider language, family care, health needs and support networks. These facts vary by destination and personal situation, so they need current local verification rather than an assumption drawn from an index. Choose one requirement that appears repeatedly in postings but is not yet clear in your experience. Test it with a small project using a relevant work product, or ask an employer contact how that task is handled and what evidence they expect from applicants. Set a threshold and review date in advance: for example, whether several credible roles meet your salary and authorization constraints and whether you can demonstrate the recurring skill gap. This is an information-gathering experiment, not proof that a destination will hire you. The OECD report “Artificial Intelligence and the Changing Demand for Skills in the Labour Market” analyzes online vacancies across ten countries and workplace comparisons. Its results illuminate skill patterns, while its scope and vacancy channel limit what it can say about any one worker or location. Keep the options proportionate to what you learn. If the task map points to a skill gap addressable inside your current work, an upgrade may preserve experience and involve less disruption. If postings repeatedly ask for adjacent responsibilities that fit your background, test that transition before assuming a larger career change is necessary. Relocation deserves more investigation only when the target work, requirements and personal feasibility all hold up. If evidence is thin, keep gathering it; no AIPI score can fill that gap.

Sources: How is AI changing the way workers perform their jobs and the skills they require?, OECD; Artificial Intelligence and the Changing Demand for Skills in the Labour Market, OECD

Choose the next evidence step, not a score-driven move

Do not move because of the IMF AI Preparedness Index. Map one recurring task, compare openings in locations you could consider, and decide whether to upgrade your current role, test an adjacent role, or investigate relocation further. The IMF’s AIPI documentation calls it an indicative national measure, warns against ranking with it, and records coverage of 174 countries with annual index data as of 2023. It can frame questions about broad enabling conditions, but cannot settle a personal move. Choose one work product you handle repeatedly and compare a manageable sample of current postings for the same target role in two feasible places. Note recurring requirements, such as a credential, language, domain knowledge, or tool workflow, and select one gap to test with a small project. Check whether authorization, salary needs, housing, health, and family obligations make either option workable. Postings are an imperfect window into hiring, so treat this as evidence gathering, not a promise of a job. Destination-specific openings and practical requirements could favor one place; national readiness may help distinguish otherwise comparable options. The free checker at /ai-job-risk-checker provides task-level change-pressure signals, not a validated probability of displacement. The paid career roadmap compares scenarios against your preferences and constraints; it does not guarantee employment or income. Use either to organize next questions, not as a relocation verdict.

Sources: AI Preparedness Index (AIPI), IMF

Questions readers ask