The DOI matches its Crossref record; 84 of the 85 references it cites were found in a registry or an index; 5 of the 84 sources its claims cite were checked for support.
Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation
Open accessAbstract
This research examines differences in consumer responses to large language model AI chatbots versus search engines when reviewing and evaluating product information. The findings suggest that consumers prefer search engines over AI chatbots and are more willing to self-disclose their personal information to search engines due to perceived familiarity. However, consumers evaluate AI chatbot-based results as less biased than traditional search engine results.
Overview
CoChat Summary
TL;DR
In an online experiment comparing search engine versus AI chatbot results for a laptop search, consumers preferred search engines and were more willing to self-disclose personal information to them, driven by greater perceived familiarity. However, relationship quality moderated these effects, and consumers rated AI chatbot results as less biased than search engine results.
Key findings
- Participants reported greater familiarity with search engine results (M = 5.58, SD = 1.33) than AI-generated results (M = 5.13, SD = 1.47; F(1, 208) = 5.63, p = 0.02, ηp² = 0.032).
- Search preference was higher for the search engine (M = 3.50, SD = 0.95) than the AI chatbot (M = 3.24, SD = 1.09; F(1, 208) = 5.23, p = 0.023, ηp² = 0.03), and self-disclosure intention was higher for the search engine (M = 3.01, SD = 1.12) than the AI chatbot (M = 2.71, SD = 1.22; F(1, 208) = 3.78, p = 0.02, ηp² = 0.03).
- Familiarity mediated both effects: indirect effect on search preference = −0.13 (95% CI: −0.255, −0.028) and on self-disclosure = −0.14 (95% CI: −0.266, −0.030).
- Relationship quality moderated the mediation for search preference (95% CI: −0.099, −0.004; familiarity × relationship quality β = 0.07, SE = 0.03, t = 2.19, p = 0.03) and for self-disclosure (95% CI: −0.133, −0.015; β = 0.13, SE = 0.03, t = 3.48, p < 0.001). AI chatbot results were perceived as less biased than search engine results (M AI = 2.28, SD = 0.75 vs. M search engine = 2.54, SD = 0.87; F(1, 209) = 5.28, p = 0.02, ηp² = 0.03). All six hypotheses (H1–H6) were supported.
Methods
- Quantitative experiment: participants imagined searching "buy a laptop" online and were randomly assigned to view either search engine results or AI chatbot results, with identical text and images across conditions and only the interface format differing.
- Online survey distributed to students (N = 260) for partial course credit; those reporting attention of "very little" or "not at all" were excluded, leaving N = 211. Laptop purchase plan in the next year served as a covariate.
- One-way ANOVAs tested main effects of information source on familiarity, search preference, self-disclosure intention, and bias perception; Process Model 4 tested mediation via familiarity and Process Model 15 tested moderated mediation with relationship quality as moderator (5000 bootstrap CIs). Harman's single-factor test was used to rule out common method bias.
Limitations
- Stated: the sample focused on young adults (Gen Z) and students, may represent particular social classes, and skewed female, reducing response variability; the authors note Druckman and Kam's argument that student and non-student samples do not differ on many factors.
- Stated: the controlled experiment used a static search engine results page versus a non-interactive AI chatbot page, a trade-off between internal and external validity that may limit real-world generalizability.
- Stated: familiarity and self-disclosure were measured with single items; the authors defend this for unidimensional, well-defined constructs but recommend multi-item measures in future research.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
5 Valid · 2 Not valid · 54 Not checked · 60 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“This increased efficiency is likely to increase consumers' evaluations of chatbot-based results as more fair and objective [70,71].”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 2.4. Sources of Information Search, Persuasion Knowledge, and Consumer Perceptions of Bias
What the sources say
- Partly supportedcites [70] Li, 2020, Research on Artificial Intelligence Customer Service on Consumer Attitude and Its Impact during Online Shopping, J. Phys. Conf. Ser. sourceExperimental — accuracy not yet measured
Our copy supports the evaluative half of the claim: Section 4.2 (Reasons for Consumers Accepting AI Customer Service) reports that 56.5% of consumers believe AI customer service 'will be more objective and neutral in the choice of information and will not deliberately choose the information that consumers like to boost sales', and the Abstract and Conclusion repeat 'absolutely neutral, more objective' as a driver of acceptance. But the paper does not derive that fairness/objectivity judgement from increased efficiency: it treats responsiveness (24/7 availability) and selectivity/objectivity as separate inherent advantages, and its efficiency evidence is quoted from another work in the Literature Review (Luo et al., a highly structured AI system 'as efficient as a senior human customer service agent and had four times the efficiency of a less experienced human agent'), where it is paired with the caution that customer emotions 'are usually determined by their subjective perceptions rather than based on objective facts'. Section 4.2 further states that accuracy/comprehensiveness are NOT the main factors driving acceptance. So the causal link the citing sentence draws (efficiency -> perceived fairness/objectivity) is not what this source says. (LLM, conf=0.68)
- Not checked — no full text of the sourcecites [71] Wang, R., Harper, F.M., and Zhu, H. (2020, January 25–30). Factors Influencing Perceived Fairness in Algorithmic Decision-Making: Algorithm Outcomes, Development Procedures, and Individual Differences. Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, Honolulu HI, USA. source
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Further, the benefits of providing complete and direct responses better fulfill the consumer's need for convenience via time-saving or the ease of accessing information for their purchases [4].”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 2.4. Sources of Information Search, Persuasion Knowledge, and Consumer Perceptions of Bias
What the sources say
- Supportedcites [4] Spatharioti, S.E., Rothschild, D.M., Goldstein, D.G., and Hofman, J.M. (2023). Comparing Traditional and LLM-Based Search for Consumer Choice: A Randomized Experiment. arXiv. source
The Discussion and Conclusion (p. n/a) states that LLM-based search 'allowed people to reach decisions faster and in fewer steps by issuing queries and receiving responses that more directly addressed the decisions at hand,' tying complete/direct responses to time savings in a consumer purchase-search task. The Introduction adds that LLM-based search 'offers a natural language interface than can handle complex queries and return detailed, direct responses,' extracting and synthesizing details across many references, and that it provides 'an easier-to-use interface that speeds up complex tasks... or free up time for them to acquire more information.' Results (4.1) confirm the time reduction (almost half the task time). Only the phrase 'need for convenience' is the citing authors' own framing, not a construct the source names; the attributed content (direct responses → saved time and easier access to purchase information) is present. (LLM, conf=0.85)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Single-item measures of familiarity are well-established and have been used in consumer behavior and psychology research due to their efficiency and validity [42,85,89,90].”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 5.3. Limitations
What the sources say
- Not checked — no full text of the sourcecites [42] Janiszewski, 1993, Preattentive Mere Exposure Effects, J. Consum. Res. source
- Not checked — no full text of the sourcecites [85] Bernardic, U., and Scheibehenne, B. (2021, January 27–30). Familiarity Attracts Consumer Attention: Two Methods to Objectively Measure Consumer Brand Familiarity. Proceedings of the 34th Bled eConference Digital Support from Crisis to Progressive Change, University of Maribor, Maribor, Slovenia.
- Partly supportedcites [89] Park, 1981, Familiarity and Its Impact on Consumer Decision Biases and Heuristics, J. Consum. Res. sourceExperimental — accuracy not yet measured
Our copy does exemplify a single-item self-report familiarity measure in consumer research: the introduction (p.1) sets out the two approaches and notes that self-assessed familiarity 'is based on the person's selfreport of how much s/he knows about the product,' and EXPERIMENTAL TASK AND DATA records a five-point very-familiar-to-unfamiliar rating used as a check on the assigned groups (DISCUSSION, means 1.89/3.13/4.05). However, the paper nowhere calls such measures well-established, offers no efficiency or validity rationale, and does not extend the point to psychology research generally; on the contrary, Construct of Product Familiarity warns that 'different individuals may employ different criteria in assessing familiarity, thus making comparisons across subjects difficult,' and the DISCUSSION states 'there is no commonly accepted (or even proposed) conceptual definition of product familiarity.' So the use-in-consumer-research part is supported while the 'well-established ... due to their efficiency and validity' characterization is not. (LLM, conf=0.72)
- Not checked — no full text of the sourcecites [90] Zhou, 2007, How Do Enhanced and Unique Features Affect New Product Preference? The Moderating Role of Product Familiarity, J. Acad. Mark. Sci. source
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The findings suggest that consumers prefer search engines over AI chatbots and are more willing to self-disclose their personal information to search engines (vs.AI chatbots) due to perceived familiarity.”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > Abstract
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“and are more willing to self-disclose their personal information to search engines (vs.AI chatbots) due to perceived familiarity.”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > Abstract
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
38 more claims
Create a free account to see all 121 claims and to verify the 6 references still open.
- Probably not a substantial claim. 62
- It may state two separate claims as one. 38
- The quote only partly states this claim. 4
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“For instance, lab experiments often manipulate the design of the assistant or give them more freedom, but they measure the results as intentions or simulated choice instead of actual checkout/payment [7,8,24,38,43,52,54,63].”
From From Recommendations to Delegation: A Systematic Review Mapping Agentic AI in E-Commerce and Its Consumer Effects (2026) · cites this paper as [54]
Experimental — accuracy not yet measured
Statistics that don’t add up
Each reported p-value recomputed from its test statistic and degrees of freedom.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“F (1, 208) = 3.78, p = 0.02”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > Self-disclosure intention.
Reported p = .02 · recomputed p = .053
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.
- Doesn’t add upF(1, 208) = 3.78reported p = .02 · recomputed p = .053
“F (1, 208) = 3.78, p = .02”
- ConsistentF(1, 208) = 5.63reported p = .02 · recomputed p = .019
“F (1, 208) = 5.63, p = .02”
- ConsistentF(1, 208) = 5.23reported p = .023 · recomputed p = .023
“F (1, 208) = 5.23, p = .023”
- ConsistentF(1, 209) = 5.28reported p = .02 · recomputed p = .023
“F (1, 209) = 5.28, p = .02”
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
“| Familiarity (Med) | 0.03 (0.05) | 0.17 | 0.34 | | Laptop purchase plan (Cov) | 0.12 (0.05) | 0.03 | 0.22 | | Indirect mediation effect (H2) | -0.13 (0.06) | -0.26 | -0.03 |”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 5.4. Recommended Future Research
Reported -0.13 · recomputed -0.0153
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > Familiarity perception.
“participants were more familiar with the search engine results (M = 5.58, SD = 1.33) than the AI-generated results (M = 5.13, SD = 1.47”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > Participants and Procedure
“as well as how familiar they were with the results shown (1 = unfamiliar; 5 = familiar).”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“(β = 0.13, SE = 0.03, t = 3.48, p < 0.001)”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > Moderated mediations.
Reported 3.48 · recomputed 4.333333
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 5.4. Recommended Future Research
“| Familiarity (Med) | 0.03 (0.05) |”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 5.4. Recommended Future Research
“| 0.17 | 0.34 |”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 5.4. Recommended Future Research
“| Laptop purchase plan | 0.13 (0.07) |”
§ Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation > 5.4. Recommended Future Research
“| -0.01 | 0.03 |”
Citations
Cites
Perceived usefulness, perceived ease of use, and user acceptance of information technology
Davis, F. D. (1989) · MIS Quarterly · doi:10.2307/249008 · Cited by 68,483 · cited 1×
“In doing so, this work sheds new light on the psychological processes of consumer engagement with AI chatbots versus traditional search engines and contributes to the literature on consumer–brand relationships, relationship marketing, and the technology acceptance model [19].”
Consumers and Their Brands: Developing Relationship Theory in Consumer Research
Fournier, S. (1998) · Journal of Consumer Research · doi:10.1086/209515 · Cited by 7,542 · cited 5×
“Similar to human relationship quality with interpersonal others, consumer–brand relationship quality indicates the strength and depth of the relationship between the consumer and the brand [13,14].”
InfluentialAttitudinal Effects of Mere Exposure
Zajonc, R. B. (1968) · Journal of Personality and Social Psychology · doi:10.1037/h0025848 · Cited by 6,745 · cited 6×
“According to Zajonc [11], an individual’s repeated exposure to a stimulus enhances their attitude toward it (e.g., greater liking) due to increased familiarity with that stimulus, which is referred to as the mere exposure effect.”
InfluentialMeasuring the involvement construct
Zaichkowsky, J. L. (1985) · Journal of Consumer Research · doi:10.1086/208520 · Cited by 6,013 · cited 1×
“Following prior work, we included the participants’ likelihood to purchase a laptop in the next year as a covariate due to its influence on their attention to and evaluations of the laptop search results [76].”
Dimensions of Brand Personality
Aaker, J. L. (1997) · Journal of Marketing Research · doi:10.1177/002224379703400304 · Cited by 4,033 · cited 1×
“For example, communicating transparency about their search algorithms or ongoing efforts to provide accurate and efficient results could increase consumers’ sincerity perception, which refers to showing positive intentions toward consumers [80], and thus facilitate strong relationships.”
The Persuasion Knowledge Model: How People Cope with Persuasion Attempts
Friestad, M., & Wright, P. (1994) · Journal of Consumer Research · doi:10.1086/209380 · Cited by 3,811 · cited 4×
“Finally, building on the persuasion knowledge model [15], consumers may recognize search engines’ commercial aspects, which can activate persuasion knowledge—consumers’ knowledge and beliefs about marketers’ persuasion goals and their underlying motives and tactics—leading to greater perceived bias…”
InfluentialInstructional manipulation checks: Detecting satisficing to increase statistical power
Oppenheimer, D. M., Meyvis, T., & Davidenko, N. (2009) · Journal of Experimental Social Psychology · doi:10.1016/j.jesp.2009.03.009 · Cited by 3,493 · cited 1×
“Among the participants, those who indicated their attention level as “very little” or “not at all” were excluded from the analysis [74], leaving a final sample of N = 211.”
Consumer engagement in a virtual brand community: An exploratory analysis
Brodie, R. J., Ilic, A., Juric, B., & Hollebeek, L. (2013) · Journal of Business Research · doi:10.1016/j.jbusres.2011.07.029 · Cited by 3,221 · cited 1×
“…constructs, such as relationship quality, and self-disclosure have been explored in the context of social media platforms—especially consumer interactions with influencers and brands [78,79]—our research extends these psychological and relational frameworks to consumer–technology interactions.”
The Predictive Validity of Multiple-Item versus Single-Item Measures of the Same Constructs
Bergkvist, L., & Rossiter, J. R. (2007) · Journal of Marketing Research · doi:10.1509/jmkr.44.2.175 · Cited by 3,107 · cited 1×
“Given this narrow and well-defined conceptualization, the use of a single-item measure aligns with prior research, which supports the appropriateness of single-item measures when a research interest center on a specific and well-defined dimension of a broader construct [87].”
Exposure and affect: Overview and meta-analysis of research, 1968–1987.
Bornstein, R. F. (1989) · Psychological Bulletin · doi:10.1037/0033-2909.106.2.265 · Cited by 1,970 · cited 1×
“In general, individuals prefer the familiar to the unknown because more risks are involved about the unknown [49,50].”
Research works printed but not linked: 2
- Ghose, A., Impeirotis, P., and Li, B. (2012). Surviving Social Media Overload: Predicting Consumer Footprints on Product Search Engines. Workshop Inf. Syst. Econ. (WISE), 1–33.
- Derlega, V.J., Metts, S., Petronio, S., and Margulis, S.T. (1993). Self-Disclosure, Sage Publications, Inc.
Cited by
Meng, K., & Xiao, J. J. (2025) · Journal of Retailing and Consumer Services · doi:10.1016/j.jretconser.2025.104496 · Cited by 13
Chatbot-aided product purchases among Generation Z: the role of personality traits
Liu, J., & Chen, J. (2025) · Frontiers in Psychology · doi:10.3389/fpsyg.2025.1454197 · Cited by 10
Balaskas, S. (2026) · Information · doi:10.3390/info17030222 · Cited by 8
Mitigating Identity Threats Posed by Artificial Intelligence: The Role of Perceived AI Relatedness
Kuchmaner, C. A. (2025) · Journal of Consumer Behaviour · doi:10.1002/cb.70066 · Cited by 6
Sun, J., & Si, J. (2025) · Journal of theoretical and applied electronic commerce research · doi:10.3390/jtaer20040341 · Cited by 4
“Conversely, some studies have shown that AI agents are perceived as more objective and fairer than human agents, which can enhance consumer trust and foster prosocial behaviors [11–13].”
Metadata
- Authors
- Kim, S., & Priluck, R.
- Affiliations
- Pace University (US)
- Corresponding author
- Kim, Soyoung
- Venue
- Journal of Theoretical and Applied Electronic Commerce Research
- Publisher
- MDPI AG
- Published
- 2025-05-05
- Volume
- 20
- Issue
- 2
- Pages
- 93
- ISSN
- 0718-1876 (ISSN-L)
- OpenAlex
- W4410081533
- Cited by
- 20
- References
- 85
- Citations by year
- 2026: 15 · 2025: 5
- FWCI
- 7.99
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Journal of theoretical and applied electronic commerce research (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Social Sciences › Business, Management and Accounting › Marketing › AI in Service Interactions
- Social Sciences › Business, Management and Accounting › Marketing › Digital Marketing and Social Media
- Social Sciences › Decision Sciences › Information Systems and Management › Technology Adoption and User Behaviour
- Keywords
- generative AI chatbots, search engines, product evaluation, information search, mere exposure effect, consumer-brand relationships, self-disclosure, perceived familiarity, search result bias, purchase decision, purchase decision process
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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation. https://app.cochat.ai/papers/10.3390/jtaer20020093
- The paper
- Kim, S., & Priluck, R. (2025). Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation. Journal of Theoretical and Applied Electronic Commerce Research. https://doi.org/10.3390/jtaer20020093
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.