Human live-streaming anchors generally inspire more trust and purchase intent than AI counterparts, but AI anchors win over efficiency-minded shoppers: intimacy drives trust for hedonic consumers, while responsiveness converts utilitarian buyers only under specific conditions.
Background AI anchors are increasingly deployed in live-streaming commerce, raising the question of whether they can substitute for human anchors. Prior studies have documented differences in consumer responses to these anchor types, but the psychological processes underlying trust formation remain unclear. This study approaches the question from a media psychology and human-machine communication perspective rather than focusing solely on commercial outcomes. Methods A between-subjects experimental design was employed. Participants ( N = 439) were randomly assigned to watch a live-streaming sales video hosted by either a human anchor or an AI anchor. Participants then completed measures of perceived intimacy, perceived responsiveness, trust, purchase intention, and motivational orientations (hedonic and utilitarian). A six-factor confirmatory factor analysis confirmed the measurement model, and moderated mediation analyses were conducted with heteroscedasticity-consistent standard errors. Results Human anchors generated higher trust and purchase intention overall. Anchor type influenced trust through two asymmetric identity-based cue pathways. Perceived intimacy (a relational cue) mediated the effect of anchor type on trust, particularly when hedonic motivation was moderate to high. Perceived responsiveness (a functional cue) did not function as a general mediator; it became a significant pathway favoring AI anchors only when utilitarian motivation was high. At low utilitarian motivation, this pathway reversed direction. Conclusion Consumer trust in live-streaming commerce is a conditional, motivation-dependent process rather than a uniform preference for either anchor type. Human anchors build trust through a broadly effective relational pathway, while AI anchors’ functional advantage converts into trust only under specific motivational conditions. These findings suggest that AI anchors will not broadly replace human anchors, but can be strategically effective when matched to efficiency-oriented consumer goals.
Summary
Main Finding
Human live-streaming anchors produce higher trust and purchase intention on average, but trust is formed via two asymmetric, motivation-dependent pathways. Perceived intimacy (a relational cue) consistently mediates the human-anchor advantage—especially when consumers are hedonic-motivated—whereas perceived responsiveness (a functional cue) becomes an AI-anchor advantage only when consumers have high utilitarian motivation (and can reverse at low utilitarian motivation). Thus AI anchors are not a blanket substitute for human anchors but can be strategically effective for efficiency-oriented audiences.
Key Points
- Overall outcome: Human anchors > AI anchors on trust and purchase intention (H1 supported).
- Two identity-based cue pathways:
- Relational pathway: perceived intimacy → trust. Stronger and broadly effective for human anchors, especially under moderate-to-high hedonic motivation.
- Functional pathway: perceived responsiveness → trust. Not a general mediator; favors AI anchors only when utilitarian motivation is high; reverses at low utilitarian motivation.
- Consumer motivational orientation (hedonic vs. utilitarian) moderates which cue drives trust.
- Theoretical framing: reframes the human-vs-AI replacement debate as a human–machine communication / media-psychology question (what consumers infer about agent identity), invoking CASA and algorithm-aversion literatures.
- Practical takeaway: match anchor type to consumer motivation and product/context rather than pursue blanket anthropomorphism or full replacement.
Data & Methods
- Design: Between-subjects experiment.
- Sample: N = 439 participants randomly assigned to watch a live-stream sales video hosted by either a human anchor or an AI anchor.
- Measures: perceived intimacy (relational cue), perceived responsiveness (functional cue), trust, purchase intention, hedonic motivation, utilitarian motivation.
- Measurement validation: six-factor confirmatory factor analysis supported the measurement model.
- Analyses: moderated mediation models tested with heteroscedasticity-consistent standard errors.
- Key empirical patterns: mediation of anchor-type → trust by perceived intimacy (hedonic-moderated); perceived responsiveness-mediated pathway significant for AI anchors only at high utilitarian motivation (and reversed at low utilitarian).
Implications for AI Economics
- Labor substitution is conditional, not universal:
- AI anchors can reduce costs (24/7 operation, lower marginal staffing costs), but cost savings must be weighed against trust and conversion losses for hedonic or relational purchase occasions.
- Expect partial substitution concentrated in product categories and time windows where consumers prioritize efficiency (e.g., commodity, utilitarian goods, late-night sessions).
- Segmentation and targeting:
- Firms should deploy AI anchors strategically to segments or contexts with high utilitarian motivation (e.g., technical product demos, price-sensitive shoppers) and retain human anchors for hedonic/brand-building contexts (e.g., fashion, cosmetics, entertainment).
- Product-market fit and platform strategy:
- Optimal adoption involves portfolio decisions: maintain a mixed anchor workforce and dynamically route viewers to anchor type based on inferred or self-reported motivation.
- A/B tests and real-time recommendation systems can personalize anchor assignment, improving ROI on AI deployment.
- Design and investment signals:
- Investments in AI should prioritize functional responsiveness (fast, accurate answers, seamless transaction flows) rather than only anthropomorphic realism, to capture utilitarian-motivated consumers.
- For hedonic contexts, investments in synthetic sociality (affective expression) may be costly and still inferior to humans; returns here are uncertain.
- Economic research directions:
- Quantify break-even points where AI cost savings outweigh trust-induced revenue losses across products and consumer segments.
- Model dynamic competition between human and AI anchors (pricing, scheduling, matching algorithms), including complementarities (e.g., AI handles routine queries; humans handle high-value/relational interactions).
- Study long-run brand and market effects (trust erosion or accumulation) from shifting anchor mixes, and cross-cultural variation in algorithm aversion.
- Policy and labor-market note:
- Expect heterogeneous displacement risk: lower for roles requiring relational trust and high for repetitive, efficiency-driven hosting tasks. Firms and policymakers should consider retraining and role reallocation strategies.
Limitations to keep in mind: lab-style experiment based on viewing manipulated videos (ecological validity), sample context not globally specified (authors based in China), and measures are self-report—further field experiments and economic modeling needed to estimate real-world revenue and labor impacts.
Assessment
Claims (10)
| Claim | Direction | Confidence | Outcome | Details |
|---|---|---|---|---|
| Human anchors generated higher trust overall than AI anchors. Consumer Welfare | positive | high | trust |
n=439
0.6
|
| Human anchors generated higher purchase intention overall than AI anchors. Adoption Rate | positive | high | purchase intention |
n=439
0.6
|
| Perceived intimacy (a relational cue) mediated the effect of anchor type on trust. Consumer Welfare | positive | high | trust (mediated by perceived intimacy) |
n=439
0.6
|
| The mediating effect of perceived intimacy on the anchor type → trust relationship was stronger (i.e., particularly operative) when participants' hedonic motivation was moderate to high. Consumer Welfare | positive | high | trust (conditional mediation by perceived intimacy moderated by hedonic motivation) |
n=439
0.6
|
| Perceived responsiveness (a functional cue) did not function as a general mediator of anchor type on trust. Consumer Welfare | null_result | high | trust (mediator: perceived responsiveness) |
n=439
0.6
|
| Perceived responsiveness became a significant pathway favoring AI anchors only when utilitarian motivation was high; at low utilitarian motivation, this pathway reversed direction. Consumer Welfare | mixed | high | trust (conditional mediation by perceived responsiveness moderated by utilitarian motivation) |
n=439
0.6
|
| Consumer trust in live-streaming commerce is a conditional, motivation-dependent process rather than a uniform preference for either anchor type. Consumer Welfare | mixed | high | trust |
n=439
0.6
|
| Human anchors build trust through a broadly effective relational pathway (perceived intimacy), while AI anchors' functional advantage converts into trust only under specific motivational conditions (high utilitarian motivation). Consumer Welfare | mixed | high | trust (mediated by intimacy for human anchors; by responsiveness for AI anchors under high utilitarian motivation) |
n=439
0.6
|
| AI anchors will not broadly replace human anchors, but can be strategically effective when matched to efficiency‑oriented (utilitarian) consumer goals. Adoption Rate | mixed | medium | market outcome implication (replacement/adoption effectiveness) |
n=439
0.06
|
| A six-factor confirmatory factor analysis confirmed the measurement model used for perceived intimacy, perceived responsiveness, trust, purchase intention, hedonic motivation, and utilitarian motivation. Other | positive | high | measurement validity (factor structure) |
n=439
0.6
|