■ CRITICAL RISK ■ Sales & Marketing
Yes, and the phone survey itself is dying alongside the job. Between online panels, robocall-grade AI voice agents, and response rates in the low single digits, paying a human to read a script into a headset is a business model on life support.
“AI surveys don't get hung up on. Well, they do, but they don't care.”
Our AI replacement risk score — how we score jobs
Telephone interviewers sit in call centers dialing through sample lists — often via automated dialers — reading standardized questionnaires for market research firms, political pollsters, and government surveys. The craft, such as it is, lies in keeping respondents on the line: coaxing someone through a 15-minute questionnaire, probing vague answers without leading them, coding open-ended responses, and following skip logic without breaking rhythm. It is scripted, monitored, metric-driven work, which is precisely the profile automation loves.
The disruption here is a pincer movement. On one side, survey research has migrated online and to SMS, where a web form costs nearly nothing per complete and doesn't need staffing in three shifts. On the other, conversational AI voice agents can now conduct the phone interviews that remain — reading questions naturally, handling 'what do you mean by that?', recording and auto-coding responses, and dialing infinitely without burnout, turnover, or supervisors listening in for quality. CATI centers that once employed hundreds of interviewers have been shrinking for years; AI voice finishes the job.
The residue of human advantage is thin. Certain populations — elderly respondents, sensitive government surveys, hard-to-reach demographic quotas — still respond better to a patient human voice, and some research designs require trained interviewers for depth interviews rather than fixed questionnaires. Qualitative interviewing, where you genuinely improvise follow-ups, is a different and safer occupation. But the standardized phone interview, which is what this job overwhelmingly is, offers almost nothing a machine can't replicate at a fraction of the cost, which is why our risk score sits at 90.
Automatability: our editorial assessment of current and near-term AI capability
Already well underway. Online panels gutted phone research over the past decade, and the surviving CATI work is now meeting AI voice agents that read scripts, log answers, and never quit mid-shift. Expect the remaining call-center interviewer seats to keep disappearing through the late 2020s, with humans retained mainly for hard-to-reach populations and government surveys that mandate live interviewers.
Yes, and they were shrinking before AI arrived. Response rates to phone surveys collapsed, pushing research online, and the CATI centers that remain are automating with voice agents that conduct scripted interviews at near-zero marginal cost. Employment in this role has been declining for years, and the trend is accelerating rather than stabilizing.
For standardized questionnaires, yes. Modern voice agents read questions naturally, follow skip logic, rephrase on request, record answers, and auto-code responses. Where they still stumble is genuine rapport — persuading a skeptical respondent to give 15 minutes — and improvised probing in qualitative interviews. Those are the shrinking islands where humans retain an edge.
More than you'd think. Structured listening, neutral probing, data-entry discipline, and the ability to keep strangers talking transfer directly into qualitative research, UX research support, panel recruitment, fundraising, and customer success. The move that pays best is toward research analysis: people who can design and interpret surveys are far safer than people who administer them.
The field is fine; the phone bank isn't. Demand for consumer insight keeps growing, but the labor shifts from data collection — which automates — to research design, analysis, and storytelling, which don't automate nearly as fast. If you like the industry, climb toward the analyst and researcher roles and treat interviewing as a foot in the door, not a destination.