What to Enter When Your Job Title Misses Your Real Duties

What should I enter when my title misses my duties?

Enter the closest truthful occupation, then use recurring duties to refine the match wherever the checker allows. If it accepts only a title, choose the profile whose work most resembles yours, note important omissions, and read the result as a proxy rather than a personal forecast. Describe present work, not a hoped-for future role. O*NET’s Job Duties Custom List asks for a current occupation title, then uses more than 19,000 occupation-specific task statements to find occupations performing similar work. This illustrates title-led duty matching; it does not show that an AI checker uses O*NET or calculates an individual exposure score. If your employer’s title is unusual, use a familiar label as an index and judge the candidate by its duties. Note meaningful omissions, such as client exceptions or approval responsibility, when the input cannot capture them. The ILO brief “Workers’ exposure to AI: What indicators tell us – and what they don’t” says exposure measures rely on static task lists and indicate technological susceptibility, not employment outcomes. A checker result can prompt closer examination; it cannot settle what will happen to your job.

Sources: Job Duties Custom List; Workers’ exposure to AI: What indicators tell us – and what they don’t

What exactly is the checker asking you to represent?

Before entering anything, identify what the field asks you to represent. It may be a searchable occupation, a set of tasks, free text, or one occupation from which the checker produces a result. These formats preserve different amounts of detail. A title field gives the system a category to look up; it does not say which parts of that occupation you perform. Do not assume a tool offers task-level controls unless you can see them. A job title is the label your employer or profession uses. A standardized occupation is a category that groups work under a shared name and description. A task statement describes a particular activity. Your task bundle is the collection of activities you actually perform, including how they connect. Your individual role also includes local procedures, authority, tools, and accountability. These layers overlap but are not interchangeable. An internal title may not match a standard category, while one occupation label can cover different tasks. The O*NET Summary Report shows how an occupational taxonomy makes some of these layers visible. Its documented profile can include an occupation code, title and definition, reported job titles, occupation-specific tasks, broader work activities, detailed work activities, and work context. O*NET distinguishes tasks specific to an occupation from work activities summarizing kinds of tasks across occupations. This can help compare a title with the work attached to a candidate profile. But it remains occupational information, not one worker’s weekly record or a measure of personal exposure. The input format sets the limit on what you can infer. With a searchable occupation field, the title is a lookup key: inspect the description before accepting a match. With a task list, selected statements may represent more of your work, if the checker lets you choose them. Free text can preserve combinations hidden by a standard label, but the result depends on how the tool interprets it. With a title-only score, the input says least about your role, so treat the output as an occupation-level proxy. No format alone establishes what your employer has adopted or whether a task will be displaced. O*NET’s Job Duties Custom List is a useful lookup example, not evidence about this checker’s design. Its page says users begin with their current occupation title and use job duties to find occupations performing similar work, drawing on more than 19,000 occupation-specific task statements. A title can start a search; duties help assess whether a resulting profile resembles the work in question. This does not mean an AI exposure checker uses O*NET, accepts its inputs, or calculates personal results in the same way. Read the field and its instructions. Are you supplying a category, choosing tasks, describing work, or receiving a score tied to a category? Interpret the result within that boundary. If the form reduces your role to one standardized label, keep your actual duties and any mismatch in view when deciding what to investigate next.

Sources: O*NET OnLine Help: Summary Report; Job Duties Custom List

How can a title help without defining your work?

An official title can be a useful starting point even when it poorly summarizes your week. Treat it as an index into occupational descriptions, not proof that one fits. If a checker asks for a title or occupation code, enter the closest recognized match you can identify, then inspect its duties before deciding whether the result represents your work. Match recurring outputs and responsibilities, not the prestige, seniority, or wording of your employer’s label. O*NET offers a concrete example of this lookup sequence. Its Job Duties Custom List asks for the current occupation title to begin a search, then uses more than 19,000 occupation-specific task statements to help find occupations that perform similar work. That is a way to search occupational information by duties; it is not an AI exposure checker, and it does not establish what another checker does with a title. A label can locate candidates; task descriptions give you something meaningful to compare. [Job Duties Custom List](https://www.onetonline.org/search/task/) Suppose, as an illustration, your employer calls you a “client solutions lead.” The phrase may lack a standardized match or point to several kinds of work. Instead of choosing whichever candidate sounds most senior, identify what you regularly produce: perhaps client reports, interpretations of unusual account results, and coordinated approvals before a change. Ask which profile best describes those outputs and actions. If one matches the title but mainly describes sales activity, while another has a less familiar title but better matches recurring analysis and coordination, the second may be a more useful comparison for this task-level question. This is an illustration, not a worker case. O*NET’s Summary Report makes the comparison inspectable. It distinguishes occupation-specific Tasks from broader Work Activities and Detailed Work Activities, and lists the occupation’s code, title, definition, sample reported job titles, and Work Context. Related Occupations can also be explored through similarities in what people do, know, and are called. A profile may miss local procedures, tools, relationships, or approval authority. [O*NET OnLine Help: Summary Report](https://www.onetonline.org/help/online/summary) Read several candidate task descriptions and ask: “Would this describe work I actually do in a normal period?” Prefer the profile that covers central recurring outputs, then note a material duty it misses. Hybrid roles may combine work that the taxonomy organizes separately. If the checker allows an alternate profile or secondary occupation, use that option for a genuine second component. If it accepts only one, keep the mismatch in your notes rather than forcing another title into the field. A title is lookup metadata, not your work history. An unfamiliar employer title does not need to be presented as though it were a recognized occupation. Start with it if the search requires it, compare candidate duties, and carry the remaining mismatch into your interpretation. The quality of the match depends on what the profile says you do, not how closely its title resembles your badge.

Sources: Job Duties Custom List; O*NET OnLine Help: Summary Report

How do you match duties without overfitting a title?

Match candidate occupations against outputs and decisions in a normal work period, then note important duties the profile misses. For each task, record what you produce or decide, the action you take, how often it recurs, and what is at stake if it is wrong or late. Start with work products, not title language. “Own client success” might mean preparing account summaries, answering routine questions, resolving unusual billing issues, or approving contract changes. Those are distinct activities under one broad label. Rewrite responsibilities as plain verb-object phrases: reconcile a report, explain an exception, approve a change, coordinate a handoff. Compare each phrase with candidate task descriptions. Ask whether a profile reflects the action and result, not just whether its title sounds familiar. O*NET’s Summary Report separates occupation-specific “Tasks” from broader “Work Activities” and “Detailed Work Activities” that can span occupations. This supports a two-level check: does a profile describe the specific result you handle, and does it capture the general kind of work? A broad activity match can point toward a candidate, but a task description that reflects your output is a closer fit. It is a structured U.S. reference, not a record of your employer’s assignments. Frequency helps identify the role’s center of gravity, but it is not the only dimension. Separate work that recurs in an ordinary week or month from duties that arise during an escalation, approval, or unusual case. Rare work can still matter when it carries accountability or a consequential decision. Preparing a standard summary may happen often, while authorizing a correction happens rarely; omit the latter and the profile may look more representative than it is. Label both routine and occasional work instead of inventing exact percentages. Imagine a hybrid client-solutions role titled “strategic partner lead.” Its outputs include account reports, explanations of unusual results, and coordination of approvals before changes. A sales profile might fit relationship work but miss report preparation; an analyst profile might fit reporting and investigation but omit client explanation and approval. It shows why the closest title or one matching task may not identify the best profile. Choose the candidate that fairly represents the assessed work and note what it misses. A job description can help recall formal responsibilities, especially when daily work has drifted from the title. Use it as a cross-check: it may bundle activities or describe expected scope rather than current work. Compare it with recent outputs and decisions. If two profiles fit, record which outputs support each and what responsibility remains unmatched. Use secondary occupations, task controls, or free-text notes only if the checker offers them. If it accepts one profile, choose the closer match and keep the gap note. The occupation is an index for comparison, not proof that every duty fits.

Sources: O*NET OnLine Help: Summary Report; Occupational Outlook Handbook

What if the checker accepts tasks or free text?

Describe work as actions and outputs, not as a broad responsibility such as “manage accounts.” If the checker offers task choices or a free-text field, use those controls to show what you do and where review or judgment enters. A compact pattern is: action + input + output + review or decision. For example: “Prepare monthly account summaries from CRM records; investigate unusual balances with clients; coordinate approval before account changes.” It does not assume the checker scores each activity separately; the form may combine them into one result. Keep activities separate when they involve different inputs, outputs, or responsibility. “Draft a response to a routine request” differs from “approve a response that changes a customer’s terms.” A tool might help prepare language, while approval carries authority and may require context unavailable in the draft. Collecting figures and interpreting an unexpected pattern are also different steps. Routing a request to the right queue differs from resolving an exception, which may require investigation and a decision. These distinctions help a reviewer see where assistance could apply and where accountable review remains, even if the checker provides only one text box. Use frequency to decide which duties to describe first, but do not turn it into a made-up score weight. Start with tasks that recur in an ordinary work period, then add a less frequent task if its consequences or responsibility matter. Do not claim an exact share of your week unless you have a reliable record and the checker asks. “Weekly reporting; occasional client escalations; final approval for account changes” is more honest than invented percentages. If entries are limited, prioritize recurring outputs and preserve consequential exceptions in a note. Research methods also distinguish tasks within occupations, though that does not establish how this particular checker works. The International Labour Organization’s “Generative AI and Jobs: A Refined Global Index of Occupational Exposure” describes an index built from task-level data, worker input, expert discussions, and AI predictions. It says most occupations contain tasks requiring human input; its modeled exposure does not mean every task under a title is equally exposed. The OECD’s “The OECD AI Exposure Measure” maps capability indicators across nine cognitive, social, and physical domains to occupational requirements. Its abstract distinguishes routine, codifiable work from contextual judgment, interpersonal understanding, complex decisions, and responsibility, while saying actual effects depend on adoption and organizational change. These studies support inspecting task differences; they do not validate an individual checker score or predict a worker’s outcome. When unsure how much detail to enter, write one line per distinct action, then combine only steps sharing the same input, output, and review. Aim for a faithful description another person would recognize, not a persuasive case for a high or low result. This gives a task-based tool or reviewer enough structure to assess potential assistance while keeping a draft, recommendation, and accountable decision distinct.

Sources: Generative AI and Jobs: A Refined Global Index of Occupational Exposure; The OECD AI Exposure Measure

Why can one occupation contain unlike exposure signals?

An occupation-level result is an average-like summary of a bundle, not a description of every task inside it. A digital, repeatable output may be technically easier for an AI system to draft or classify than a task that depends on local context, interpreting an exception, maintaining a relationship, or taking accountable approval. If signals are mixed, ask which task changes and who verifies or owns the result. The International Labour Organization’s 2025 study, “Generative AI and Jobs: A Refined Global Index of Occupational Exposure,” makes task variation central to its method. Using a representative sample of 29,753 tasks in Poland’s occupational classification, a survey of 1,640 employed people across ISCO-08 major groups, and 52,558 data points on perceived automation potential for 2,861 tasks, the researchers compared results with a survey and Delphi-style discussions among international experts, then used an AI assistant to predict scores for tasks in ISCO-08 documentation. The resulting framework reports four progressively increasing exposure gradients. Tasks in one occupation therefore need not share a signal; gradients represent modeled exposure, not an all-or-nothing boundary. The ILO page reports that one in four workers globally are in occupations with some generative-AI exposure and that 3.3% of global employment is in its highest gradient. These are employment shares mapped to occupational exposure categories, not estimates that those workers will lose their jobs. It says most occupations include tasks requiring human input and describes transformation as more likely. That supports examining workflow change; it does not show human involvement prevents automation or preserves headcount. Review and responsibility may shift. A separate lens appears in the OECD working paper “The OECD AI Exposure Measure.” It maps AI capabilities across nine cognitive, social, and physical domains to occupational requirements. The abstract calls it forward-looking and updateable: routine or codifiable activities differ from contextual judgment, interpersonal understanding, complex decisions, and responsibility. The OECD also states that actual effects depend on adoption, regulation, organizational change, and social choice. Capability is only one link: data may be unavailable, checking costly, or adoption absent. Observed use alone does not establish redesigned work, changed demand, or displacement. Do not compare an ILO gradient directly with an OECD measure as if they were readings on one shared scale: their task definitions, methods, scope, and time horizons differ. Instead, use each as evidence that occupation-level labels compress varied work and that exposure is not an employment outcome. Treat human contributions as prompts for inspection, not as proof of safety. For one task, name its input, output, review burden, and accountable decision-maker. Ask what changes if AI assists production: who checks facts, resolves exceptions, explains decisions, or carries responsibility? An occupation-level signal can therefore help locate questions for review, but it cannot substitute for describing the work that actually fills a week.

Sources: Generative AI and Jobs: A Refined Global Index of Occupational Exposure; The OECD AI Exposure Measure

What can an exposure result tell you, and what can it not?

An exposure result indicates modeled overlap between assessed AI capabilities and occupational tasks. It cannot show that a tool reliably does your work, that your employer adopted it, or that your job will be eliminated. Treat it as a prompt to investigate, not a personal risk percentage or forecast. The word exposure often compresses distinct claims.

Capability is what a system may do under an assessment's conditions and measures. Applicability asks whether it fits a task's inputs, output, and operating conditions. A system able to draft text may be relevant to drafting; that does not establish it can produce an acceptable result from information available at a particular workplace. Reliability, error costs, privacy rules, and human checking affect usefulness. These are workflow questions, not conclusions contained in an occupational label.

Observed use means evidence that people in a defined sample used a tool for specified work. Its meaning depends on who was studied, when, and how use was measured. Task compatibility is not evidence of use. Employer adoption is different: an organization's decision to permit, buy, integrate, or require a tool, shaped by security, regulation, cost, systems, and management. The OECD's AI Exposure Measure is a forward-looking occupational measure that identifies adoption, regulation, organizational change, and social choice as factors shaping actual effects. That caveat limits what it establishes.

Adoption still does not settle what happens to a job. Workflow redesign changes responsibilities or task sequences: a tool might draft an output for a worker to check, or change who handles routine steps and exceptions. Labor demand concerns how many workers or skills employers seek; displacement concerns workers losing work or jobs. Neither follows automatically from exposure or adoption. Redesign claims need process evidence, demand claims need local labor-market evidence over time, and displacement claims need employment outcomes credibly linked to technology rather than other changes.

The ILO brief Workers’ Exposure to AI: What Indicators Tell Us—and What They Don’t says exposure indicators describe technological susceptibility, not labor-market outcomes, and may omit economic and institutional constraints. A checker can still flag a task for review without predicting later links in the chain. A high signal alone cannot tell you whether an employer changed workflows, workers use a system, demand shifted, or anyone was displaced.

When you see a result, ask what evidence answers the next question. For current use, seek documented use in a relevant setting and check its similarity to your work. For adoption, rely on confirmed policies, approved tools, or observed changes rather than assumptions about plans. For labor demand, consult separate local employment evidence; exposure is not a hiring outlook. Keep these records distinct from the checker result. Its practical role is narrower: identify a task whose inputs, review burden, or accountability deserve attention, then investigate what is happening around it.

Sources: Workers’ exposure to AI: What indicators tell us – and what they don’t; The OECD AI Exposure Measure

What should you do when the tool accepts only a title?

If the checker accepts only one occupation title, enter the nearest honest match based on the work you regularly do. Keep a note of duties, task shares, exceptions, or approval responsibilities the result cannot represent. Do not rename your job to chase a lower or higher signal. The result is a proxy for a broader occupational profile, not a personal forecast. Treat that limitation as a boundary on the result, not a reason to abandon the exercise. Record the entered title and why it fits; your usual outputs; recurring duties; and any less frequent task with meaningful consequences. Add who reviews the work and relevant local context, such as tools or procedures in use. Keep the note factual: a documented workflow is useful context, while a guess about what management may adopt is not evidence. That record creates a second line of inquiry. The checker can flag modeled overlap between AI capabilities and occupational tasks. Workplace evidence answers a different question: whether a tool is currently approved or used for a particular step, what review is required, and whether a workflow has actually changed. Look for policy, documented process, or a direct description of current practice. Do not treat a capability signal as proof that your employer has adopted a system, or treat rumors and vendor claims as confirmation of a plan. The International Labour Organization’s “Workers’ exposure to AI: What indicators tell us – and what they don’t” explains that exposure measures rely on static task lists, vary by method, and leave out adoption constraints such as economic conditions and institutional barriers. The brief says these indicators describe technological susceptibility, not labor-market outcomes: they do not establish whether firms find automation feasible or how employment, wages, and demand will adjust. A title-only checker therefore has two limits at once: the profile may not fit your exact role, and even a good task match cannot answer what your organization will do. O*NET’s “Summary Report” help page illustrates what an occupational profile can and cannot preserve. It lists occupation-specific tasks separately from broader work activities and work context, alongside titles and other descriptors. That structure is useful for checking why a profile resembles your work, but it remains an occupational description rather than your personal weekly record. Your gap note carries the missing context forward; it does not turn a broad result into a precise individual measurement. If you compare results later, keep the entered title and note together, and compare only like-for-like inputs and methods. A score can point you toward a task worth examining, but its precision cannot exceed the information the form accepted. Start with one recurring or consequential duty and verify how it is done, checked, and currently supported. That gives you a concrete question for a small workflow review without turning a screening signal into a career verdict.

Sources: Workers’ exposure to AI: What indicators tell us – and what they don’t; O*NET OnLine Help: Summary Report

What if no occupation matches, or your work crosses countries?

If no occupation matches cleanly, use the closest local classification you can find by comparing duties and outputs, then state what the match leaves out. If your work crosses countries, do not translate your title word for word and assume it names the same job everywhere. Codes, title conventions, and role scope differ across systems. A crosswalk locates candidates; it does not make national job descriptions identical. Start with the country or labor market your question concerns. Search its official occupational database by duties: outputs, decisions, and recurring exceptions. Compare two or three plausible descriptions for overlap in actual work. If the checker uses a different country’s taxonomy, select the nearest profile only as a proxy and keep a short note with the mismatch. If it accepts task descriptions, use those controls to represent the duties directly; don’t imply that every checker supports them. O*NET’s Job Duties Custom List offers one example of a duty-based lookup in the United States. The interface asks for a current occupation title as a starting point, then says its database uses more than 19,000 occupation-specific task statements to find occupations performing similar work. That can help someone with an unfamiliar internal title find candidates. It remains a U.S. occupational information system, though, and its search is not an AI exposure score or a substitute for a local classification. The U.S. Bureau of Labor Statistics’ Occupational Outlook Handbook also describes occupations and reports information such as education, pay, and projected outlook for the U.S. labor market. Those fields do not measure technical exposure to AI. For international context, the International Labour Organization’s “Generative AI and Jobs: A Refined Global Index of Occupational Exposure” describes a method built from task data, expert input, and model predictions. Its underlying task sample came from Poland’s occupational classification, with survey responses from 1,640 employed people across broad ISCO-08 groups and 52,558 judgments covering 2,861 tasks. The paper then uses an AI assistant to predict scores for tasks in ISCO-08 documentation. This shows how task evidence can connect with an international classification; it does not establish that a local checker uses that mapping or predicts adoption in your country. The study concerns potential generative-AI exposure, not your employer’s choices or an individual displacement outcome. Keep local demand evidence separate from exposure evidence. A national outlook source may describe employment projections or training requirements; an exposure measure estimates overlap between task descriptions and assessed capabilities. Neither substitutes for the other; U.S. projections are not forecasts for another country. When no profile captures your hybrid role, choose the closest candidate by recurring work, name the major omitted duty or accountability, and treat the result as directional task evidence. The uncertainty belongs beside the match, not hidden behind a precise-looking score.

Sources: Job Duties Custom List; Occupational Outlook Handbook; Generative AI and Jobs: A Refined Global Index of Occupational Exposure

How should you turn the result into a proportionate next move?

Choose one recurring or consequential task the result raises. Note the input, output, reviewer, errors and where approval enters. An exposure signal can prompt inquiry; it cannot establish whether your workplace uses AI for that task or whether change would improve work. The ILO brief “Workers’ Exposure to AI: What Indicators Tell Us—and What They Don’t” says exposure indicators describe technological susceptibility, not employment, wage, or demand outcomes. The OECD’s “The OECD AI Exposure Measure” also separates adoption and organizational change. Seek evidence rather than assuming an employer’s plans.

Check policy and data permissions first. Compare one permitted output with the usual process; review errors and checking time. If the task is changing, learn for a specific gap, such as evaluating generated summaries, rather than taking a generic course because a score is high. Compare staying and redesigning the work, an adjacent move, and a larger change against your salary floor, location, training time, health, and family responsibilities. Start by listing one normal week’s task groups and investigating one output. The free checker at /ai-job-risk-checker provides task-level change-pressure signals; the paid roadmap compares scenarios, without guaranteeing employment or income. Treat either as a prompt for inquiry, not a career verdict.

Sources: Workers’ exposure to AI: What indicators tell us – and what they don’t; The OECD AI Exposure Measure

Questions readers ask