Why part of this work moves and part of it doesn’t
This job is built on two very different activities. One is routine: read questions from a prepared script, key the answers into a form, and check the record for blanks or contradictions. Software handles that shape of work well. The other is social: find the right person, explain why the survey or intake matters, and keep them talking long enough to finish. That part is persuasion, not data entry.
Survey and intake interviewers also work inside rules. Sampling plans, consent wording and privacy requirements decide who may be contacted and how. A call that drifts off script can invalidate a response or breach a condition of the study. So the question of will AI replace interviewers is really a question about who is accountable when an answer is recorded wrong.
The labor market is already moving without any dramatic hand-off. The Bureau of Labor Statistics counts about 148,060 people in this occupation and projects employment falling 10.4% between 2025 and 2035 (BLS, 2025). Median pay is $45,920 a year (BLS, 2025). Online panels, text surveys and self-service forms took volume long before conversational AI arrived.
What software runs, what it supports, and what stays with a person
Start with the tasks closest to pure process. Asking fixed-wording questions in order and transcribing replies into a database sits in the group AI can run on its own, along with flagging responses that are missing or inconsistent. Our task split puts 26% of task time there. How we measure that share is explained on the coverage method page.
The assist group is larger than people expect. Scheduling and rescheduling contacts, dialing and routing calls, and preparing clean summaries of completed interviews all go faster with a tool in the loop while a person still decides what counts as done. That slice is 65% of task time.
Then there is the residue that keeps the job a job: convincing a reluctant respondent to take part, judging whether an answer is honest or confused, and adapting to someone who is distressed, hard of hearing, or speaking through a relative. By our count that group is 9% of task time. It is small in hours and heavy in consequence, because a bad answer poisons the data.
Good to know: this occupation covers survey, market research and intake interviewers, not the hiring managers who interview job candidates.
What has actually been tested
No one has run a head-to-head test of AI against trained interviewers on live respondents, which is why the evidence grade is D. A grade at that level means we publish no quality parity number for this job, because none has been measured in a way we trust.
What would settle it is specific: a field study comparing response rates, break-off rates and data quality between human-administered and AI-administered interviews on the same sample, with the same questionnaire and the same consent script. Until that exists, claims in either direction are marketing. You can read how we grade evidence on our methodology page, and see the same grading applied across jobs in the full rankings.
When the picture could shift
Most likely between 2035 and 2045 (8 in 10 of our scenarios). What that window means, and how it is built, is set out on the replacement-year method page.
Two things could pull it earlier. First, cost: this work needs no robotics at all, so adoption is a software decision, not a capital project. Second, volume pressure: when a research firm needs thousands of short, standard interviews, automated calling and chat is an obvious place to cut hours.
Two things hold it back. Consent, privacy and data-quality rules mean a client often has to approve an automated instrument before it touches a sample, and that approval is slow. And refusal behavior matters: people hang up on machines more readily, and a rising break-off rate destroys the savings. A cheaper interview that nobody completes is not cheaper.
How to stay needed
Lean into the work that software cannot carry. Recruiting and retaining hard-to-reach respondents is the most valuable skill in the field. So is reading an answer for truthfulness rather than accepting it, and handling sensitive subjects such as health, income or household conflict without pushing someone away.
Two skills are worth adding. One is survey and data quality: understanding sampling, weighting basics, and how to spot a corrupted batch of responses. The other is tool supervision, which means running an automated instrument, checking its transcripts, and correcting what it mislabels. That turns you into the person who signs off, not the person being replaced in a workflow.
Nearby roles are worth a look if you want to move sideways. Eligibility interviewers for government programs apply rules to real cases and carry more decision authority. Loan interviewers and clerks work in regulated finance with documented checks. Human resources assistants use the same interviewing and record-keeping skills inside an employer rather than a research firm.
For context on where this sits, browse information and record clerks, the wider call center sector, or our list of jobs expected to shrink. If you are weighing a move, put this role and your target role side by side in the job comparison tool.