The conversation around AI and job displacement tends to swing between two extremes: either everything will be automated tomorrow, or nothing will really change. Neither is accurate. To cut through the noise, we built a model that evaluates over 800 occupations listed in the Bureau of Labor Statistics database, scoring each one based on the proportion of its core tasks that current and near-future AI systems can realistically perform. The result is a nuanced risk map that reveals which roles are genuinely vulnerable and which are more resilient than headlines suggest.
Our analysis breaks each occupation into its component tasks and evaluates them against five automation dimensions: data processing volume, decision-rule complexity, physical environment variability, interpersonal interaction depth, and creative originality requirements. Roles that score high on routine data processing and low on physical variability and interpersonal depth sit at the top of the risk spectrum. This includes many back-office administrative positions, basic data entry roles, and certain categories of financial analysis where the work follows well-defined patterns. Importantly, it is not entire occupations that get automated, but specific tasks within them. A financial analyst who spends 70% of their time pulling and formatting reports faces very different pressures than one who spends 70% of their time advising clients through complex restructuring decisions.
On the other end of the spectrum, roles with high physical variability, deep interpersonal interaction, or significant creative demands remain substantially harder to automate. Skilled trades like electricians and plumbers operate in unpredictable physical environments that current robotics cannot navigate reliably. Healthcare professionals who combine clinical judgment with patient communication sit in a similar position. And roles that require genuine creative synthesis, not just pattern remixing but original strategic thinking, continue to resist automation in meaningful ways. The takeaway is not that these jobs are immune, but that their task profiles make full automation significantly further out on the timeline.
One of the most important findings is that job title alone is a poor predictor of risk. Two people with the same title at different companies can have wildly different task distributions. A marketing manager at a small firm who writes copy, manages campaigns, and talks to customers daily has a different risk profile than one at a large enterprise who primarily reviews dashboards and approves templated content. This is why we designed our assessment tool to ask about specific daily tasks rather than just job titles. The granularity matters because it determines whether your plan should focus on upskilling within your current role or pivoting to an adjacent one with a better long-term outlook.
The data also reveals a middle ground that most commentary ignores: roles where AI will significantly change the work without eliminating it. These transformation-zone occupations, which represent roughly 40% of the jobs we analyzed, will see major shifts in how work gets done. Professionals in these roles have a window of opportunity to adapt, learn to work alongside AI tools, and reposition themselves as the human layer that makes automated systems actually useful. The workers who recognize this window and act on it will be in a far stronger position than those who wait for the change to arrive.