An AI voice interview is a software agent that asks hiring questions by phone or voice call and scores the conversation. A work sample is a role-specific task whose output a hiring manager can open, read, and defend in a debrief. Candidate satisfaction with the voice screen measures experience. It does not measure whether the person can do the job.
Bullhorn’s 2026 GRID Talent Trends report (23 September 2026) surveyed nearly 2,300 staffing candidates. Among those interviewed by an AI voice agent, 92% rated the experience as good as or better than a human interview — 62% said better (up from 88% overall parity in the prior wave). About 57% of respondents had completed an AI voice interview. Separately, 93% of candidates with a positive AI experience said they would use that staffing firm again. Those numbers are real and useful for employer brand. They are not a substitute for an openable artifact of day-one work. A high satisfaction score tells you the screen felt fair and fast. It does not tell a hiring manager what the candidate would ship on Tuesday.
This guide is for lean TA and hiring managers who keep hearing “candidates love our AI voice agent” in vendor decks. It separates experience metrics from hiring evidence, and shows where Dotof AI Role-Simulation Assessments fit: prove the work first, then spend live conversation — human or voice — on what the task could not answer. For the broader interviewer-vs-simulation buy decision, see AI interviewer vs job simulation.
What does Bullhorn’s 92% actually measure?
It measures how candidates felt about an AI voice interview relative to a human one — speed, clarity, and whether the interaction felt as good as or better than talking to a person. GRID’s design is a staffing-talent pulse, not a predictive-validity study. The headline does not claim that the voice score predicts job performance, reduces bad hires, or replaces a work sample. Treat it as a candidate-experience signal for firms that run high-volume screens after hours. Celebrate the brand upside. Do not import it into a debrief as proof of capability.
Firms that ignore satisfaction still hurt themselves: GRID also found that nearly all candidates with a positive AI touchpoint would reuse the firm. Experience matters for offer acceptance and agency loyalty. The mistake is collapsing “they liked the call” into “they can do the role.” Those are different questions with different artifacts.
Why isn’t a high satisfaction score enough for the hiring manager?
Because the hiring manager’s job in a debrief is to open evidence of work — not a sentiment bar. A voice transcript and a fit score answer “how did they talk about the job?” A work sample answers “what would they produce?” Managers defend shortlists to peers and executives with the second answer. A polite, fast AI call can leave the packet empty: no triage notes, no customer reply, no ranked pipeline, no plan. When the only artifact is a conversation score, the panel rediscovers capability live — which is the expensive screen you thought you automated.
Satisfaction can even mask a weak signal. Candidates may prefer a voice agent that is easier to schedule and less intimidating than a panel, while still producing answers that do not predict day-one output. Experience and validity can move in opposite directions. Keep both metrics on the dashboard; do not let one replace the other.
When should HR use a work sample instead of (or after) an AI voice interview?
- Use the voice screen when scheduling is the bottleneck, questions are structured, volume is high, and a human still owns the advance/reject decision. Keep disclosure clear: what is scored, what is not, and that a person decides.
- Add a work sample before panel time when the role has observable day-one deliverables, a wrong hire is costly, or managers refuse to interview without something they can open. Generate the task from the live JD so it is not a generic quiz — see from job description to role simulation.
- Prefer the sample over the voice score as the shortlist rank for professional roles where “can they do the work?” is the question you cannot afford to get wrong. Use the voice call after — or beside — the task for surge coverage, not as the sole early filter.
Do not ask the simulation and the voice agent the same questions. Rewrite the conversation around judgment, motivation, and gaps the work sample left open. Keep a written human override: no model score should be the only vote. The practical gates are in the responsible AI hiring checklist.
What should the hiring manager open after the AI voice screen?
An openable packet: the task brief, the candidate’s work, the rubric, and a short note on integrity conditions. That is the anatomy of a useful AI assessment report for hiring managers — not a single fit percentage from a conversation. If your voice vendor only returns a transcript and a score, you still need a role task before scarce panel hours. Pair integrity controls so the sample is trustworthy under the conditions it was taken; see how to prevent AI cheating in hiring assessments.
How does Dotof fit this decision?
Dotof AI Role-Simulation Assessments start from the job description, generate a realistic task, run it in a supervised browser, and return structured evidence hiring managers can compare. Use an AI voice interview if you need after-hours question coverage and a strong candidate experience. Use the simulation when the decision you cannot afford to get wrong is whether the person can do the job. Satisfaction and proof can coexist — they should not be confused.