AVGPAY RESEARCH CENTER · AI AND WORK

Will AI change your job? Start with the evidence.

Search 825 occupations to explore task exposure, what the job actually involves, how much it pays and what workforce projections say. A high AI task-exposure score is not a predicted chance of losing your job.

What the workforce research shows

AvgPay combines the historic task-exposure study with published wage and job information. These lenses answer different questions and should not be mistaken for a single predictive AI replacement model.

33/100Worker-weighted national model score
46.4%Workers in our high or very-high exposure bands
22Career fields compared
2025–2035BLS employment projection window in the dataset

Exposure measures 2023-era potential task acceleration. These bands are AvgPay editorial categories, not calibrated unemployment risk. The BLS outlook is a separate forecast reflecting multiple economic and technological assumptions.

Explore charts and the full research →

Which career fields look most exposed?

These are worker-weighted study-based task scores by occupation group. Explore individual careers before drawing conclusions about employment or long-term security.

  1. Computers & math65/100
    21 occupation categories covered
  2. Office & admin57/100
    54 occupation categories covered
  3. Business & finance53/100
    32 occupation categories covered
  4. Engineering48/100
    36 occupation categories covered
  5. Science47/100
    48 occupation categories covered
  6. Legal47/100
    8 occupation categories covered

Five questions worth asking before a career change

These original cross-dataset research guides combine wages, measured exposure and job outlook. They explain the selection rules, show the underlying occupations and outline what the numbers do not establish.

Browse the full research library →

What the risk bands actually mean

  • Low0–14Few tasks met the study’s hypothetical AI speedup criteria.
  • Moderate15–34Some tasks met the study’s criteria. Actual workplace change may differ.
  • High35–59Many tasks met the study’s criteria; this is not a job-loss forecast.
  • Very high60+A large share of tasks met the study’s potential speedup criteria, not proof of automation.

The thresholds are AvgPay presentation choices, not scientifically calibrated probabilities. They classify modeled task exposure under historical assumptions, rather than real-world job losses.

How we connect the evidence

Task exposure: an external study estimated which job tasks could be accelerated using AI. Pay: occupational wages describe the distribution across people currently employed. Outlook: BLS projects decade-long employment and openings, independently of our exposure bands. Work and skills: occupation profiles identify the work to investigate, not individually validated automated tasks.

Because no one measure can predict your future, the value comes from looking for agreement—and disagreement—among these signals. A well-paid occupation with high exposure can still grow; a low-exposure occupation can still face falling demand for unrelated reasons.

Original exposure research: Eloundou, T., Manning, S., Mishkin, P. and Rock, D. GPTs are GPTs: Labor market impact potential of LLMs. Science (2024); working paper arXiv:2303.10130 (2023). Study. AvgPay methods and limitations.