The exact text we saved

This is the version exactly as it was saved, character for character. The SHA-256 hash below is a 64-character code worked out from that text. Run the same text through SHA-256 and you get the same code; change a single letter and the code comes out completely different. Each version also stores the previous version’s hash, which chains them together so no version can be swapped out quietly.

TRUVACE RECORD VERSION
record: TRV-2026-2653
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-12T16:56:59.622754Z
status: published
lens: g_space
sector: business
headline: Emotional Resonance or Rational Guidance? Exploring the Mechanisms Through Which Generative AI Recommendation Roles Shape Consumers' Adoption Intentions
dek: ABSTRACT Generative artificial intelligence (GenAI) is reshaping e‐commerce by enabling conversational recommenders that can explain their advice and flexibly enact social roles (e.g., assistant vs. friend). Yet it remains unclear how these roles affect consumers' adoption of GenAI recommendations, through which trust pathways, and under what message designs. Drawing primarily on role theory and trust theory, with product evaluability and dual‐process logic specifying the relevant conditions, we examine how reco…
gain_title: In a three-experiment study (n = 263, 293 and 367), assistant-style AI recommendations produced higher adoption intentions for search goods and friend-style recommendations produced higher adoption intentions for experience goods.
problem_title: (none)
trace_subject: (none)
gain_reading: In a three-experiment study (n = 263, 293 and 367), assistant-style AI recommendations produced higher adoption intentions for search goods and friend-style recommendations produced higher adoption intentions for experience goods.
gain_evidence: assistant‐type recommendations generate higher adoption intentions for search goods, whereas friend‐type recommendations generate higher adoption intentions for experience goods
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers ran three experiments on how a generative AI recommender's social role affects whether consumers say they would adopt its advice. In Study 1 (n = 263), assistant-type recommendations produced higher adoption intentions for search goods, while friend-type recommendations did so for experience goods.

Study 2 (n = 293) traced the pathways: assistant-type advice raised adoption of search goods mainly through cognitive trust, and friend-type advice raised adoption of experience goods mainly through affective trust.

Study 3 (n = 367) tested message framing. Loss framing strengthened the assistant and search-good pairing, and gain framing strengthened the friend and experience-good pairing. The authors suggest e-commerce firms align AI roles and framing with how easily a product can be evaluated.
limitation: 
tag: Evidence-backed gain
key_points: Study 1 (n = 263) found a role and product fit: assistant-type recommendations raised adoption intentions for search goods. | In Study 1, friend-type recommendations raised adoption intentions for experience goods. | Study 2 (n = 293) found that assistant-type advice worked mainly through cognitive trust for search goods. | Study 2 found that friend-type advice worked mainly through affective trust for experience goods. | Study 3 (n = 367) found loss framing strengthened the assistant and search-good fit, while gain framing strengthened the friend and experience-good fit. | The authors describe role plasticity as the same AI system's capacity to enact different socially meaningful roles.
rundown: The work draws on role theory and trust theory, with product evaluability and dual-process logic used to specify when each role should work. It looks at how recommender role interacts with product type and message framing.

Search goods are products whose quality can be judged before purchase, and experience goods are judged through use. Assistant-style advice matched the first kind and friend-style advice matched the second, according to the three experiments.

The authors conclude that role-contingent trust and context-dependent message processing explain how the same AI system persuades differently, and they offer guidance on pairing roles and framing strategies in e-commerce.
sources:
- peer_reviewed | Journal of Consumer Behaviour | https://doi.org/10.1002/cb.70243 | 2026-09-03
prev: 0000000000000000000000000000000000000000000000000000000000000000

SHA-256 hash

954f9be0e1d49499204b3b1ff33f1c764591b24b17a929e092b0ab14012a714a

Previous version’s hash

0000000000000000000000000000000000000000000000000000000000000000