CoChat

Examining Humanness as a Metaphor to Design Voice User Interfaces

Open access
Desai, Smit, Dubiel, Mateusz, Leiva, Luis A. · 2024 · ACM Conversational User Interfaces 2024

The DOI matches its Crossref record; 77 of the 79 references it cites were found in a registry or an index, and 2 are not checked yet; 8 of the 58 sources its claims cite were checked for support.

Abstract

Voice User Interfaces (VUIs) increasingly leverage ‘humanness’ as a foundational design metaphor, adopting roles like ‘assistants,’ ‘teachers,’ and ‘secretaries’ to foster natural interactions. Yet, this approach can sometimes misalign user trust and reinforce societal stereotypes, leading to socio-technical challenges that might impede long-term engagement. This paper explores an alternative approach to navigate these challenges—incorporating non-human metaphors in VUI design. We report on a study with 240 participants examining the effects of human versus non-human metaphors on user perceptions within health and finance domains. Results indicate a preference for the human metaphor (doctor) over the non-human (health encyclopedia) in health contexts for its perceived enjoyability and likeability. In finance, however, user perceptions do not significantly differ between human (financial advisor) and non-human (calculator) metaphors. Importantly, our research reveals that the explicit awareness of a metaphor’s use influences adoption intentions, with a marked preference for non-human metaphors when their metaphorical nature is not disclosed. These findings highlight context-specific conversation design strategies required in integrating non-human metaphors into VUI design, suggesting tradeoffs and design considerations that could enhance user engagement and adoption.

Verified 2026-10-05 · 2 references not checked yet

Overview

CoChat Summary

TL;DR

In a 2x4 mixed-subjects study with 240 frequent VUI users, human metaphors (doctor, financial advisor) were compared with non-human metaphors (health encyclopedia, calculator) in health and finance. Human metaphors were more enjoyable and likeable in health and more trusted overall, but finance showed no significant metaphor differences; awareness of a non-human metaphor affected adoption intentions.

Key findings

  • In health, the human metaphor (doctor) led to higher perceived enjoyment than the non-human metaphor (health encyclopedia), F(1, 238) = 4.663, p = .032, η²p = 0.019, and higher likeability, F(1, 238) = 7.551, p = .006, η²p = 0.031. Means: enjoyment 5.43 ± 1.12 vs. 5.32 ± 1.24; likeability 4.32 ± 0.61 vs. 4.22 ± 0.69.
  • In finance, user perceptions did not significantly differ between the human metaphor (financial advisor) and non-human metaphor (calculator) for enjoyment, likeability, trust, intelligence, or intention to adopt (e.g., enjoyment interaction p = .832; likeability p = .325; trust p = .659; intelligence p = .325; intention p = .262).
  • Overall, participants trusted human metaphors more than non-human metaphors, F(1, 238) = 5.577, p = .019, η²p = 0.023; overall enjoyment also favored human metaphors, F(1, 238) = 9.798, p = .002, η²p = 0.040, as did overall likeability, F(1, 238) = 8.913, p = .003, η²p = 0.036.
  • There was no significant overall main effect of metaphor type on intention to adopt, F(1, 238) = 1.193, p = .276, η²p = 0.005, but a significant interaction involving awareness and non-human metaphors emerged, F(1, 238) = 4.116, p = .044, η²p = 0.017, with a marked preference for non-human metaphors when their metaphorical nature was not disclosed. Perceived intelligence did not differ by metaphor across domains, overall F(1, 238) = 0.447, p = .504, η²p = 0.002.

Methods

  • The study used a 2x4 factorial design: awareness level (explicit vs. implicit) was between-subjects, and metaphor type was within-subjects with four levels—doctor and health encyclopedia in health, financial advisor and calculator in finance. Participants experienced all four metaphor types within their assigned awareness condition, with presentation order counterbalanced.
  • Participants were 240 Prolific users (M = 43.21, SD = 13.73), all U.S.-based, highly proficient English speakers with prior VUI experience; 82.5% reported being very or extremely familiar with VUIs, and 58.3% used them daily. Audio clips were generated with Speechify TTS, using a female-sounding voice for VUI “Z” and a male-sounding voice for the user.
  • Measures included perceived enjoyment, intention to adopt, trust, likeability, and intelligence, plus optional qualitative feedback. Main analysis used repeated measures ANOVA with metaphor type as the within-subjects variable and awareness as the between-subjects factor.

Limitations

  • The paper states that participants formed impressions from hypothetical dialogues rather than direct interaction, making findings observational rather than experiential; it also notes that VUI perceptions and user relationships change over time, so longitudinal research is needed.
  • The paper states that the study was restricted to native English speakers living in the U.S. because metaphor use depends on cultural factors, limiting generalizability across cultures.
  • Evident from the text: the qualitative thematic analysis was conducted by the first author alone, and the study used scripted six-turn dialogues with synthetic TTS voices, which may not capture real-world VUI interaction dynamics.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1145/3640794.3665535

Cites

77 of 79 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

8 Valid · 2 Not valid · 39 Not checked · 52 Own results and interpretations

47 more claims

Create a free account to see all 101 claims and to verify the 50 references still open.

See all claims — free account
  • Probably not a substantial claim. 37
  • It may state two separate claims as one. 19

Create a free account to see these claims.

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • ConsistentF(1, 238) = 4.663reported p = .032 · recomputed p = .032

    “F(1, 238) = 4.663, p = .032”

  • ConsistentF(1, 238) = 0.376reported p = .54 · recomputed p = .540

    “F(1, 238) = 0.376, p = .540”

  • ConsistentF(1, 238) = 0.045reported p = .832 · recomputed p = .832

    “F(1, 238) = 0.045, p = .832”

  • ConsistentF(1, 238) = 0.023reported p = .879 · recomputed p = .880

    “F(1, 238) = 0.023, p = .879”

  • ConsistentF(1, 238) = 9.798reported p = .002 · recomputed p = .002

    “F(1, 238) = 9.798, p = .002”

5 of 28 · the rest aren’t loaded

23 more recomputed tests

Create a free account to see every test the Stats check recomputed.

See all tests — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Self-check results appear here once an independent check has confirmed them.

Citations

Cites

Research works printed but not linked: 4

  1. Heloisa Candello, Claudio Pinhanez, David Millen, and Bruna Daniele Andrade. 2017. Shaping the experience of a cognitive investment adviser. In Design, User Experience, and Usability: Understanding Users and Contexts: 6th International Conference, DUXU 2017, Held as Part of HCI International 2017, Vancouver, BC, Canada, July 9-14, 2017, Proceedings, Part III 6. Springer, 594–613.
  2. Amanda Cercas Curry, Judy Robertson, and Verena Rieser. 2020. Conversational assistants and gender stereotypes: Public perceptions and desiderata for voice personas. In Proceedings of the second workshop on gender bias in natural language processing. 72–78.
  3. George Lakoff and Mark Johnson. 1980. Metaphors We Live By. University of Chicago Press. Google-Books-ID: r6nOYYtxzUoC.
  4. Don Norman. 2013. The Design Of Everyday Things (revised edition ed.). Basic Books, New York, New York.

Cited by

See all 25

Metadata

Authors
Desai, S., Dubiel, M., & Leiva, L. A.
Affiliations
  • Northeastern University (US)
  • University of Luxembourg (LU)
Venue
ACM Conversational User Interfaces 2024
Publisher
ACM
Published
2024-07-08
Pages
1-15
OpenAlex
W4400397492
Cited by
21
References
73
Citations by year
2026: 9 · 2025: 10 · 2024: 2
FWCI
6.21
Type
conference paper
Language
en
Open access
Locations
  • ACM Conference on Conversational User Interfaces (CUI) (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • Open Repository and Bibliography (University of Luxembourg) (submittedVersion, cc-by)
  • arXiv (Cornell University) (cc-by)
Funders
  • Horizon 2020 (CHIST-ERA-20-BCI-001)
  • European Innovation Council and Small and Medium-sized Enterprises Executive Agency (101071147)
Topics
  • Social Sciences › Business, Management and Accounting › Marketing › AI in Service Interactions
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
  • Physical Sciences › Computer Science › Human-Computer Interaction › Innovative Human-Technology Interaction
Keywords
voice user interface, design metaphor, user trust, social stereotypes, adoption intention, health domain, conversational design, metaphor awareness, user engagement

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-04, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-04 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.