Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation

Abstract

This research examines differences in consumer responses to large language model AI chatbots versus search engines when reviewing and evaluating product information in the information search stage of the purchase decision process. Building on the mere exposure effect, the consumer–brand relationships literature, and consumer decision-making research, this research investigates consumer search preferences and self-disclosure willingness along with a psychological mechanism. We conducted an experiment comparing two types of product search tools and tested a moderated mediation using Process Model 15. 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. However, we find that the quality of consumers’ relationship with the source of information moderates this effect. Further, we show that consumers evaluate AI chatbot-based results as less biased than traditional search engine results. Contrary to previous research that has mainly focused on AI chatbots’ functional aspects in consumer adoption, the current research emphasizes the psychological and relational aspects of product search perception and preference. In this way, we offer novel insights into a purchase decision-making process and consumer engagement as consumers adopt AI chatbot technology for product search.

1. Introduction

Generative AI burst into the consumer marketplace with the introduction of ChatGPT in November of 2022 [1]. ChatGPT is an AI chatbot comprising large language models (LLMs) pre-trained on an extensive amount of textual data [2], offering consumers a new way to search for information on the Internet [3]. Whereas traditional search engines such as Google provide lists of links, AI chatbots such as ChatGPT have a natural conversational interface, which allows for detailed and direct interactive results in response to complex queries [4]. With these advantages, the use of AI chatbots has been growing. For example, in February of 2025, Open AI reported 400 million weekly users, up 33% from December of 2024 [5]; meanwhile, on 31 October 2024, ChatGPT introduced Search GPT, recognizing the ability of AI chatbots to compete with search engine firms [1].

While the increasing popularity of AI chatbots has significantly changed consumers’ information search behavior and the retail landscape [6], research on AI Chatbots is still in its early stage [7] and has not kept pace with managerial practices, leaving empirical guidance limited. Although a growing body of research has begun to investigate AI chatbots across different contexts (e.g., e-commerce and service failure), previous research has mainly focused on single AI chatbots or comparisons among similar AI-based systems (e.g., Amazon), leaving comparative research on AI chatbots versus traditional search engines unknown [8]. However, given that traditional search engines remain a dominant search method, playing an important role in consumer decision-making [9,10], it is imperative to understand whether and how AI chatbots (e.g., ChatGPT) compared with traditional search engines shape consumers’ product search perceptions and evaluations. To this end, the current research aims to bridge the critical gap between practice and research and adds to the current literature by identifying and testing the factors that influence consumer perceptions and evaluations of AI chatbots versus a traditional search engine for product information search.

The theoretical underpinnings of our research build on (1) the mere exposure effect [11], (2) the relationship psychology [12] and consumer–brand relationships literature [13], and on (3) the persuasion knowledge model. We suggest that consumers will prefer to use and be more willing to self-disclose personal information to search engines over AI chatbots due to perceived familiarity. The mere exposure effect suggests that individuals develop more favorable attitudes toward a stimulus with repeated exposure due to increased familiarity [11]. However, we propose that this positive effect is likely to be enhanced (vs. reduced) depending on consumers’ relationship quality with search tools, which indicates the strength and depth of the relationship between the consumer and search tools [13,14]. 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 perceptions of the search engine results compared with those of AI chatbots. Bias perception refers to an inclination in favor or against a stimulus [16]. Thus, we predict that consumers will evaluate AI chatbot-based search result as less biased than those obtained with search engines.

The current research makes important theoretical contributions to the marketing and psychology literatures and provides the managerial implications.

First, while the conversational interaction features of AI chatbots (ChatGPT) disrupt the sequential search mode of traditional search engines [3], little research has empirically tested whether and how AI chatbots in comparison with traditional search engines differentially affect consumer perception and search preference. Despite the disruption, traditional search engines have been the dominant method for information search, continuing to grow at an 8% rate [10]. Moreover, given firms’ substantial investment in search engine optimization (SEO), examining the comparative effects of search engines versus newly emerging AI chatbots on consumer search preference and self-disclosure offers important and timely insights into the dynamics of consumer search behavior.

Second, by examining the mediating role of perceived familiarity and the moderating effect of consumers’ relationship quality, our research introduces new psychological and relational dimensions to consumer search behavior and adds to a growing body of research on AI chatbots. In other words, going beyond the utilitarian aspects of AI chatbots emphasized in previous research [17,18], we suggest that consumers can build quality relationships with technology, demonstrating that perceived familiarity and relationship quality with search tools matter in influencing consumer search preference. 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].

Third, this research contributes to the consumer decision-making literature. Information search is a critical step in guiding subsequent evaluations of alternatives and, ultimately, consumers’ final purchase decisions [20]. As a result, firms attempt to influence this step to be included in consumers’ consideration sets, which lead to final purchase decisions. This research shows how and why familiarity and relationship quality influence consumer product search processes and provides useful insights into understanding consumer purchase decision processes in the AI chatbot digital era.

Lastly, this research provides important implications for online search advertising and retailing. Firms have invested millions of dollars in search engine optimization strategies to achieve better visibility on the Internet and remain competitive in the marketplace [21]. As AI chatbots have shifted how consumers access information, firms should understand the impact of these behavioral changes on the effectiveness of their digital promotion strategies. We show that consumers exhibit greater preferences for and willingness to disclose their personal information with search engines than AI chatbots, despite bias perceptions toward the search engines. Thus, this work provides important practical implications that can help firms to assess, revisit, and adapt their advertising strategies. In addition, prepurchase data, such as information search behavior, help retailers understand the purchase intent of consumers to target them more effectively. As consumers have experimented with AI chatbots for product search, understanding consumer search behavior provides important implications for retail companies in formulating retail strategies [22]. More concrete guidelines and practical implications are further elaborated in Section 5.

2. Theoretical Background

2.1. Two Sources of Information Search: Search Engines and AI Chatbots

Search Engines: Search engines are widely used to locate product-specific information for purchase decisions [23]. For example, Google’s search engine is the most visited website globally, with an average use duration of over 10 min [24]. When consumers access a search engine for product research, they search in stages. Specifically, the consumer first formulates a query that represents the desired information; the search engine then returns a list of relevant links related to the consumer’s queries, based on popularity and authority, and the consumer determines whether the results produce a usable response. If not, the consumer reformulates the query until the results are satisfying [3].

A large body of work in marketing has examined consumer search behaviors using search engines. For example, previous research has found that online search engines lower the cost of locating information without affecting expectations or utility for consumers [25,26]. Researchers have suggested that good search results can save consumers money on purchases, whereas inefficient results lead to added costs. The efficiency of web search results may vary depending on the consumer’s age. While traditional search engines allow consumers to access reference material from different sources (e.g., displaying domains or publishers of different results), synthesizing information from multiple sources can be time-consuming [4]. For example, consumers must often click through several different web results and read these various web pages to find relevant information, which may generate feelings of frustration with the task of online product search. Hence, as alternatives to product search and discovery, consumers have paid attention to AI chatbots to improve their search results.

AI Chatbots: The recent introduction of AI chatbots has shifted how consumers access and search information online [27]. Unlike traditional search engines, which offer a list of links, chatbot-based search offers a natural language interface that can handle complex queries and return detailed, context-aware responses [3]. Trained on extensive text data from the Internet, AI chatbots can extract details from many different references and integrate potentially complex information across multiple sources [4]. Consumers engage in a conversational exchange to refine and follow up on a sequence of inquiries, which can reduce complexity and simplify tasks.

With the increasing real-world applications of AI chatbots, a growing body of research has begun to investigate the uses of AI chatbots across different contexts. For example, some studies have focused on the efficacy of AI chatbot applications in the e-commerce domain [6,28] or their relative effectiveness to other comparable systems. Specifically, in e-commerce, a dual expert–expert classification system—that is, a domain-specific expert trained on large e-commerce domain data and a general expert managed by AI chatbots—can accurately categorize products [28]. In making product recommendations, ChatGPT may have a significant impact on consumers’ trust in it, leading to higher consumer intention to adopt its consideration set than those of other AI recommenders (e.g., Amazon) [29].

Other researchers have examined the factors that influence the consumer perception and adoption of AI chatbots, specifically ChatGPT [30,31,32] and found that quality, accuracy, timeliness, audience familiarity, and congruence with other sources play important roles in affecting its credibility [30]. Consumers may also prefer to receive more recommendations from ChatGPT than from a human or an online travel agency and report a higher level of satisfaction [33].

While AI chatbots offer many advantages, other studies have pointed out their limitations. For example, AI chatbots may lack the ability to comprehend human speech nuances and contextually deeper meanings [30,34]. Because AI chatbots are pre-trained on data, they may provide inaccurate, outdated, and unreliable information [35,36] Kim et al. [33] have also noted that AI chatbots may have limited domain expertise, an inability to control mistakes, deficient originality, and ethical issues. Ethical concerns have been raised with the use of ChatGPT, such as the potential negative impacts of AI chatbots on genuine human connections and interactions, privacy, and human ingenuity [30,35,37].

In summary, while recent research has investigated the uses of AI chatbots (both benefits and drawbacks), the scope of research has mainly focused on their functional aspects or technical factors that affect their credibility and consumer perception, leaving the comparative roles of AI chatbots versus traditional search engines in the product search domain relatively unknown. Thus, this research explores (1) the comparative effects of AI chatbots versus traditional search engines on consumer search preference and self-disclosure intentions, along with familiarity as an underlying mechanism, and (2) the moderating role of consumers’ relationship quality with search tools.

2.2. The Mediating Role of Familiarity in Consumer Search Preference and Self-Disclosure

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. A large body of research suggests that familiarity affects individuals’ information processing, attitudes, and behavior [11,38,39]. For example, when objects are familiar, individuals tend to process them faster and more accurately than when objects are novel [39]. Familiarity has been positively associated with customer decision-making and behavior. For example, perceived familiarity with a brand increases consumer attention and leads to positive responses to that brand [39,40] with less cognitive processing of alternatives. The more familiar consumers are with a product category, the more likely that familiarity will lead to a more elaborate cognitive structure, resulting in deeper evaluations of factors [11].

Familiarity reflecting prior encounters with the same service or product indicates the recognition of reliable information about the provider’s services or products [41,42]. Thus, consumers who are familiar with a product or service are more likely to repurchase the same product or service or continue to use the same services that they have experienced. Numerous studies have supported the finding that familiarity increases customer loyalty and repeat purchases [43,44,45,46].

Search engines have been ubiquitous for consumers who perform online information searches since the 1990s, whereas AI chatbots present a new method for accessing purchase information. Thus, consumers are more familiar with and have more experience with search engines than with AI chatbots. Taken together, given consumers’ long history with and use of search engines and building on the mere exposure effect, we predict that consumers will prefer to use search engines over AI chatbots for product search and that familiarity will drive their responses:

H1. 

Consumers will exhibit a greater preference for using a search engine over an AI chatbot.

H2. 

Consumers’ preference for a search engine over an AI chatbot will be driven by familiarity perception.

Furthermore, we suggest that familiarity with search engines will increase consumers’ comfort with self-disclosure. Self-disclosure, which originates from the psychology literature [12], refers to individuals’ voluntary sharing of unobservable personal information with others [47,48]. In general, individuals prefer the familiar to the unknown because more risks are involved about the unknown [49,50]. As consumers’ familiarity with a product or service increases with repeated exposure, their knowledge or expertise about a product or service should increase. This enhanced knowledge reduces the perceived risk. Previous research has supported the inverse relationship between familiarity and reduced perceived risk [51,52]. For instance, brand familiarity reduces perceived risk in consumer purchases of online clothes [53] and intimate purchases [54]. In our context, we focus on the depth of self-disclosure—the extent to which consumers are willing to share personal and sensitive information (e.g., financial information) rather than superficial information (e.g., book interest). Given that consumers value privacy and are generally cautious about sharing personal information online [55], focusing on the depth of self-disclosure provides a more meaningful way to capture consumers’ behavioral intentions online. Bonnin [56] also suggested that increased familiarity can mitigate perceived risk in the context of augmented reality, and such reductions in perceived risk can facilitate both purchase decisions and self-disclosure. Taken together, we predict that consumers will be more willing to disclose their personal information to a search engine than an AI chatbot, due to greater familiarity with the search engine:

H3. 

Consumers will be more willing to disclose their personal or sensitive information to a search engine than an AI chatbot.

H4. 

Consumers’ self-disclosure willingness will be mediated by their familiarity with a search engine (vs. an AI chatbot).

2.3. The Role of Relationship Quality in Moderating Responses to Online Product Search

Thus far, we have suggested that familiarity positively influences consumers’ search preferences and self-disclosure intentions. However, the strength of consumers’ relationships with search tools may interact with perceived familiarity to change their search preferences and self-disclosure willingness. For instance, although AI chatbots are newer to consumers (i.e., a low level of familiarity), AI chatbots may facilitate consumer–chatbot co-creation relationships, leading to increased value for both players [57]. This consumer relationship quality may moderate the positive familiarity effect; that is, depending on the quality of the relationships that consumers may have with the source of information or search tools, the familiarity effect can be intensified or reduced.

Research on consumer–brand relationships supports this prediction [13]. The consumer–brand relationship literature suggests that consumers build relationships with inanimate objects, such as brands, as they build relationships with humans [13]. 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].

A large body of research has demonstrated that consumer–brand relationship quality positively influences consumer purchase and repurchase intentions [58,59], reluctance to switch brands, and willingness to share personal information with the company [14]. The quality of consumer–brand relationships also improves customer retention and loyalty, increasing future sales. Relationship quality also reflects stronger emotional and psychological closeness to that brand, which encompasses the subjective feeling of proximity in a relationship [59]. Thus, as consumers feel emotionally close to a brand, they are more willing to share their personal information with it.

Consumers’ relationship quality with the information sources should play a similar role. Pentina et al. [60] have found that consumers’ perceived relationship quality with an online social network encourages future intentions to continue using this network and increases preferences for other brands that utilize the same network. Along the line, Zajonc [11] suggests that while familiarity with a stimulus typically enhances individuals’ attitudes, negative experiences or relationships with the stimulus may attenuate the positive effect of familiarity. That is, if consumers build negative experiences or lower relationship quality with the source, the positive influence of familiarity is likely diminished. Hence, building on the consumer–brand relationship literature, we propose that as consumers’ relationship quality with a source of information increases, the effects of familiarity on their search preference and self-disclosure intentions for that source will increase. Therefore, we hypothesize that consumers’ relationship quality with sources of information will moderate the positive effect of familiarity on their search preference and self-disclosure intentions:

H5a. 

The relationship quality with a source of information (search engine vs. AI chatbot) will moderate the relationship between familiarity and search preference.

H5b. 

The relationship quality with a source of information (search engine vs. AI chatbot) will moderate the relationship between familiarity and self-disclosure intentions.

2.4. Sources of Information Search, Persuasion Knowledge, and Consumer Perceptions of Bias

Persuasion knowledge refers to consumers’ knowledge and beliefs about marketers’ persuasion goals and their underlying motives and tactics, as well as how persuasion operates [15]. The persuasion knowledge model [15] suggests that marketers may harbor ulterior, self-serving motivations, and consumers often know that a persuasion attempt is being directed at them [61]. Specifically, in sales and advertising settings, when persuasion knowledge is activated, the consumer’s ability to recognize, reflect on, and assess marketers’ intentions and efforts increases [62]. Marketing research suggests that commercial intent and attempts to persuade may invoke negative responses from consumers [63,64]. For example, if consumers believe that marketers employ potentially manipulative tactics to sell a product, service, or idea (i.e., the activation of persuasion knowledge), they become skeptical of the motivations and engage in coping strategies, displaying less favorable evaluations and responses to offset the commercial nature of the communication [15,63,65].

As consumers are familiar with search engines, they are likely aware of their advertising and e-commerce functions. For example, the commercial nature of an information source can be revealed in the form of sponsorships, native advertising, or product placements; paid and organic results, when clicked, lead to websites and purchase options. In addition, the results may not clearly present accurate information [66], as search engines may favor large companies that may pay to ensure their results are featured on the first page, which may generate less varied and more limited results [67,68,69]. Thus, with increased familiarity, the presence of commercial features of traditional search engines may activate consumers’ persuasion knowledge, leading to increased bias perceptions of search engine results. On the other hand, AI chatbots are nascent, and the commercial intent of these services might be less clear to consumers; thus, consumers may be less able to notice a source’s motives and be more likely to focus on the benefits to themselves in using the data [61]. 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]. This increased efficiency is likely to increase consumers’ evaluations of chatbot-based results as more fair and objective [70,71]. Taken together, building on the persuasion knowledge model, we predict that consumers may be more aware of the commercial and persuasion functions of search engines, thereby evaluating AI chatbot-based search results as less biased and more objective than those obtained from the search engines:

H6. 

Consumers will evaluate AI chatbot search as less biased than search engines.

Figure 1 displays our conceptual framework.

3. Materials and Methods

To test our hypotheses, we employed a quantitative experimental research method. Specifically, using an experimental design, we tested the main effects of the source of information (i.e., AI chatbots or search engines) on consumer search preference, self-disclosure willingness, and bias perceptions via ANOVA. Second, using Process Model 15 [72], we tested a moderated mediation model to examine whether the main effects would be mediated by familiarity perception, as well as whether a consumer’s relationship with a source of information would moderate the relationship between familiarity and search preference and self-disclosure willingness.

Participants and Procedure

An online experimental survey was distributed to students (N = 260) who completed the study for partial course credit. Student samples are commonly used in marketing along with online panels and can provide a representative sample of the general consumer population [73]. 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. Participants were told that they would participate in a study that aimed to understand how information influences individuals’ attitudes toward products. Next, the participants proceeded to a page where they were asked to imagine a situation where they entered the keywords “buy a laptop” online and were told that they would view and examine the search results on the following page. When examining the search results, participants were randomly assigned to one of two conditions: (1) search engine results vs. (2) AI chatbot results. In the search engine condition, participants viewed the laptop search results generated by a search engine and were asked to examine the same words and information. Specifically, participants were presented with three laptop models with information about the price and the image of each—a Lenovo Thinkpad X, Dell Inspiron 15 Laptop, and HP Pavilion Aero Laptop—along with examples of questions other people also ask, such as “What brand of laptop is best?”, “Which laptop is worth to buy?” (see Appendix A for stimuli). In the AI chatbot condition, participants viewed the laptop results generated by the AI chatbot, which remained constant in this condition and contained the same information as that of the search engine results. Specifically, we used identical text and images across both conditions, ensuring the only substantive variation was the interface style (i.e., search engine list format vs. conversational chatbot format). The only difference between the AI chatbot and the search engine conditions was that, in the AI chatbot condition, the information was presented in an AI chatbot format, which included the following introduction: “That’s great! Choosing a laptop can be tough, but I can help you narrow down your options for all, shopping, videos, tools, images, perspectives and more. Here are some popular options to consider:”

Then, participants answered four questions about their bias perception of the search results—the degree to which the displayed search results were impartial, unbiased, objective, and based on facts—on all 5-point Likert scales (1 = strongly disagree; 5 = strongly agree), as well as how familiar they were with the results shown (1 = unfamiliar; 5 = familiar). Participants indicated their search preferences, such as (1) their willingness to use this search information again, (2) their intention to run their next search the same way, and their self-disclosure intention, such as “I am comfortable providing my financial and personal information to the company” (1 = strongly disagree; 5 = strongly agree). Participants answered two questions measuring their relationship quality with the company presenting the source of information that was shown on the results page: “I am very committed to my relationship with the company” and “The relationship with the company is something I intend to maintain for a long period of time” [75] (1 = strongly disagree; 5 = strongly agree). They then reported their likelihood of buying a laptop in the next year (1 = very likely; 5 = very unlikely). Lastly, participants completed their attention check and answered demographic questions (see

Table 1. Demographic information of participants (N = 211).

Measure	Percentage
Age
Under 18	0.5%
18–24	93.8%
25–34	2.8%
35–44	0.5%
45–54	1.4%
55 or above	0.9%
Gender
Male	34.6%
Female	63.0%
Non-binary	0.9%
Prefer not to say	1.4%

).

4. Results

Familiarity perception. We conducted a one-way ANOVA with the source of information (search engine vs. AI chatbot) as an independent variable and perceived familiarity as a dependent variable. The results revealed that 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; F (1, 208) = 5.63, p = 0.02, ηp² = 0.032).

Search preference. Prior to the analysis, we averaged the two items—(1) the willingness to use the same information again and (2) the willingness to run next search in the same way—to create an overall search preference score ($\gamma$ = 0.68). 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]. Then, we conducted an ANOVA using the source of information (search engine vs. AI chatbot) as an independent variable, a next-year laptop purchase plan as a covariate, and search preference as a dependent variable. The results revealed the significant effect of the search engine (vs. AI chatbot) on search preference (M_{search-engine} = 3.50, SD = 0.95, M_AI = 3.24, SD = 1.09; F (1, 208) = 5.23, p = 0.023, ηp² = 0.03): participants showed a greater preference for using the search engine over the AI chatbot, confirming H1.

Self-disclosure intention. An ANOVA analysis showed the significant effect of the search engine condition (vs. the AI chatbot condition) on a self-disclosure intention (F (1, 208) = 3.78, p = 0.02). Participants indicated a greater willingness to provide their personal and financial information than those under the AI chatbot condition (M_search = 3.01, SD = 1.12, M_AI = 2.71, SD = 1.22, ηp² = 0.03), confirming H3.

Mediations. To test the effect of the search engines (vs. AI chatbots) on behavioral search preferences and self-disclosure intentions via familiarity perception, we followed Model 4 of the Process Macro [72]. A bootstrapping confidence interval for the indirect effect of the source of information on (1) search preference and (2) self-disclosure intention revealed significant mediation by familiarity perception (search preference: indirect effect = −0.13, 95% CI: −0.255, −0.028; self-disclosure: indirect effect = −0.14, 95% CI: −0.266, −0.030). That is, participants under the search engine condition were more familiar with search engine results than those under the AI chatbot condition ($\beta$ = −0.51, p = 0.01), increasing their search preference and self-disclosure willingness regarding the company ($\beta$_{search preference} = 0.25, $\beta$_{personal disclosure} = 0.22, all ps < 0.001), confirming H2 and H4.

Moderated mediations. Prior to conducting the analysis, two relationship quality items were combined into an overall relationship quality score ($\gamma$ = 0.93). To test the moderating role of relationship quality for the mediating relationships between the source of information and search preference via perceived familiarity, we used Model 15 of the Process Macro [72], with the source of information (search engine = 0 vs. AI chatbot = 1) as an independent variable, familiarity perception as a mediator, relationship quality as a moderator, laptop purchase plan as a covariate, and search preference as a dependent variable. A bootstrapping confidence interval revealed significantly moderated mediation by relationship quality for the participants’ search preference (95% CI: −0.099, −0.004). Specifically, the interaction between perceived familiarity and relationship quality in affecting search preference was significant ($\beta$= 0.07, SE = 0.03, t = 2.19, p = 0.03), indicating that the main effect of the AI chatbot (vs. the search engine) on search preference via perceived familiarity was moderated by relationship quality, confirming H5a.

Process Model 15 was again used for self-disclosure intention; a bootstrapping confidence interval revealed significantly moderated mediation by relationship quality for self-disclosure willingness (95% CI: −0.133, −0.015); the interaction between perceived familiarity and relationship quality in affecting self-disclosure willingness was significant ($\beta$= 0.13, SE = 0.03, t = 3.48, p < 0.001), which confirmed H5b.

Bias perception. Prior to the analysis, we reverse-coded and averaged four items (impartial, unbiased, objective, and based on facts) to create an overall bias score (α = 0.87). An ANOVA showed that participants under the AI chatbot condition perceived the results as less biased than those under the search engine condition (M_AI = 2.28, SD = 0.75, M_{search-engine} = 2.54, SD = 0.87, F (1, 209) = 5.28, p = 0.02, ηp² = 0.03), confirming H6 (see

Table 2. Direct effects, mediation effects, and moderated mediation effects.

	b Coefficient (SE)	Lower 95% BCBCI	Upper 95% BCBCI
Direct Effect
Outcome: Search Preference (H1)
ChatGPT vs. Search Engine	−0.32 (0.14)	−0.59	−0.04
Laptop purchase plan	0.16 (0.05)	0.07	0.26
Outcome: Self-disclosure (H3)
ChatGPT vs. Search Engine	−0.38 (0.16)	−0.69	−0.07
Laptop purchase plan	0.22 (0.06)	0.11	0.33
Mediation Effect
Outcome: Familiarity
ChatGPT vs. Search Engine (X)	−0.51 (0.20)	−0.89	−0.13
Laptop purchase plan	0.13 (0.07)	−0.01	0.03
Outcome: Search Preference
ChatGPT vs. Search Engine (X)	−0.20 (0.13)	−0.47	0.06
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
Outcome: Self-disclosure
ChatGPT vs. Search Engine (X)	−0.24 (0.15)	−0.54	0.06
Familiarity (Med)	0.27 (0.05)	0.16	0.37
Laptop purchase plan (Cov)	0.18 (0.05)	0.08	0.29
Indirect mediation effect (H4)	−0.13 (0.06)	−0.26	−0.03
Moderated Mediation Effect
Outcome: Search Preference
ChatGPT vs. Search Engine (X)	−0.55 (0.36)	−1.25	0.15
Familiarity (Med)	−0.06 (0.11)	−0.28	0.16
Relationship Quality (Mod)	−0.02 (0.22)	−0.45	0.40
XxMod	0.13 (0.11)	−0.09	0.34
MedxMod	0.07 (0.03)	0.01	0.14
Laptop purchase plan (Cov)	0.01 (0.04)	−0.07	0.09
Index of moderated mediation (H5a)	−0.04 (0.03)	−0.10	−0.004
Outcome: Self-disclosure
ChatGPT vs. Search Engine (X)	−1.33 (0.39)	−2.11	−0.55
Familiarity (Med)	−0.23 (0.12)	−0.47	0.13
Relationship Quality (Mod)	−0.39 (0.24)	−0.86	0.08
XxMod	0.37 (0.12)	0.13	0.61
MedxMod	0.13 (0.03)	0.06	0.21
Laptop purchase plan (Cov)	0.05 (0.05)	−0.05	0.14
Index of moderated mediation (H5b)	−0.07 (0.03)	−0.14	−0.02

Notes: BCBCI = bias corrected 5000 bootstrap confidence intervals, Med = mediator, Mod = moderator, Cov = covariate.

,

Table 3. Results of measurement model.

Constructs	Indicators	Factor Loadings	Composite Reliability	Average Variance Extracted (AVE)	Cronbach’s Alpha
Bias perception toward a source of information (BP)	BP1	0.84	0.89	0.67	0.87
	BP2	0.83
	BP3	0.82
	BP4	0.79
Search preference (SP)	SP1	0.74	0.70	0.53	-
	SP2	0.73
Relationship Quality (RQ)	RQ1	0.86	0.84	0.73	-
	RQ2	0.84

, and

Table 4. Correlations matrix.

Constructs	Familiarity	Search Preference	Relationship Quality	Self-Disclosure Intention	Bias Perception
Familiarity	1
Search preference	0.38 **	1
Relationship quality	0.27 **	0.56 **	1
Self-disclosure intention	0.36 **	0.68 **	0.55 **	1
Bias perception	−0.32 **	−0.49 **	−0.36 **	−0.45 **	1

Note: **: correlations are significant at the 0.01 level.

).

Common method bias. Common methods bias can arise in research conducted via self-reported measures when both independent and dependent variables are measured within one survey. To ensure that common method bias did not influence the validity of our study, we first ensured the confidentiality of participants’ responses to minimize potential socially desirable biases. The questionnaire was administered online and anonymously, restricted to a single submission per participant to maintain data reliability and confidentiality. Second, to ensure content validity, we emphasized to the participants that there were no right or wrong answers. We also checked readability across conditions; however, there was no significant difference in terms of readability between the two conditions (p > 0.45). Lastly, we applied Harman’s single-factor test. However, no single factor exceeded the threshold value of 50%; thus, the assumption of common method bias was ruled out [77].

This experimental study provides support for our hypotheses (see

Table 5. Hypotheses test results.

Hypotheses	Test Results
H1: ChatGPT (vs. Search Engine) → Search preference	Supported
H2: ChatGPT (vs. Search Engine) → Familiarity → Search preference	Supported
H3: ChatGPT (vs. Search Engine) → Self-disclosure intention	Supported
H4: ChatGPT (vs. Search Engine) → Familiarity → Self-disclosure intention	Supported
H5a: ChatGPT (vs. Search Engine) → Familiarity x Relationship Quality → Search preference	Supported
H5b: ChatGPT (vs. Search Engine) → Familiarity x Relationship Quality → Self-disclosure intention	Supported
H6: ChatGPT (vs. Search Engine) → Bias perception	Supported

). As predicted, consumers exhibited greater preferences for and willingness to share their sensitive personal information with a traditional search engine (Google) over an AI chatbot. These effects were driven by familiarity perception and moderated by consumers’ relationship quality with the source of information. However, consumers perceived the results of the AI chatbot as being less biased than those of the search engine (see Figure 2).

These findings suggest that, despite the growing use of AI chatbots, perceived familiarity influences consumer search preference and self-disclosure intentions. In addition, over and above seeking utilitarian benefits, our findings show that consumers can develop quality relationships with these search tools, which serves as a boundary condition for the familiarity effect. This finding provides important implications in terms of relationship marketing and branding strategies for search engine and AI chatbot companies, especially regarding how to build and reinforce their relationships and brand images with consumers. Lastly, our findings offer practical implications for search tool companies, regarding how they can deal with the commercial aspects of their services and communicate this aspect with consumers to mitigate bias perceptions.

5. Discussion

The introduction of AI Chatbots, such as ChatGPT on 22 November 2022, has significantly reshaped how consumers search for information on the Internet, disrupting traditional search engines like Google [3]. Concerned about how ChatGPT would affect their core business, Google issued a “code red” over the rise of ChatGPT. However, despite the rapid changes in the search industry, little research has explored how and why AI chatbots affect consumer responses to product search results and their behavioral intentions in comparison with traditional search engines. Thus, our comparative research provides important, timely contributions. Focusing on Generation Z consumers and using an experimental design via a moderated mediation model, this research demonstrates that consumers prefer traditional search engines to AI chatbots by showing their greater intention to use the same information again and a greater willingness to conduct their next search in the same way. Consumers are more willing to reveal their personal and sensitive information to a search engine firm than an AI chatbot firm. All these behavioral intentions are mediated by perceived familiarity. However, we observed a boundary condition: consumers’ relationships with the source of information moderates these mediating effects. Furthermore, the study’s results suggest that consumers evaluate search engine results as being more biased than those of AI chatbots.

5.1. Theoretical Contributions

First, this research contributes to the literature on information search. The emergence of AI chatbots represents a technological shift in the way consumers access and search information online [3]. Although the new paradigm of AI chatbots challenges traditional search models (e.g., Google’s sequential search via a list of web links), search engines have served as the primary method for information search and continue to play an important role in consumer decision-making [9]. Through comparing the impact of AI chatbots with that of search engines on consumer perception and evaluations, our research provides a more complete and deeper understanding of the dynamics of consumer information search.

Second, this research contributes to the literature on consumer–brand relationships and relationship psychology. While familiarity perception increases consumers’ search preference and self-disclosure intentions, our research shows that the strength of familiarity may vary based on the quality of consumers’ relationships with the search tools. Whereas previous research has mainly focused on the functional attributes of AI chatbots and their influence on consumer perception, we propose a novel perspective, such that consumers can also form a strong relationship with search tools. Further, while the relational 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. By doing so, this work sheds new light on the underlying relationship dynamics and psychological processes of consumer product search behavior.

Third, the current work contributes to the literature on consumer judgment and decision-making regarding product decisions. Product information search is one of the first critical steps in guiding subsequent evaluations of product alternatives, ultimately affecting final purchase decisions and post-purchase satisfaction [20]. By identifying and testing the factors that influence consumer product search preferences and self-disclosure intentions, this research highlights the importance of familiarity and quality of relationships in the consumer information search phase, providing useful findings about consumer search and purchase decision-making processes in the age of AI information search.

5.2. Managerial Implications

This work provides managerial implications. First, our findings show that consumers perceive search engine-based results as more biased. Search engine companies may wish to reduce this negative perception. This bias perception might be related to the commercial part of search engine functions, such as sponsored or pay-per-click ads that can activate consumer persuasion knowledge, invoking negative perceptions. Activating persuasion knowledge may adversely impact building quality relationships with consumers. Furthermore, the ability of AI chatbots to provide quick, complete, and direct answers to consumer queries is known to increase the perceived objectivity of the search results [71]. Thus, search engine companies may consider mitigating these biases, for example, through incorporating AI features into their search functions. In fact, Google has added AI features to its search engine, allowing consumers to see AI overviews. This may increase the fairness perception of results and help search engines overcome negative bias perceptions. In addition, while Google shares information about its Ad Rank—how its paid advertising ranking operates online—it may wish to communicate this information more explicitly to consumers, as increased awareness of this knowledge can reduce consumers’ bias perceptions and establish closer relationships with Google.

Second, the current research suggests that relationship quality is an important moderator. As AI chatbots are new and consumers are unfamiliar with this new search tool, AI chatbot providers may try to build strong relationships with their users. 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. Further, given that consumers engage in interactive conversations with ChatGPT using natural language, AI chatbot firms could emphasize the social aspects of their online search environment to elicit more dialog with consumers. To do so, AI chatbots firms could use informal and friendly language. For example, using contractions instead of full negations (e.g., “I can’t (vs. cannot) wait to hear your response to my answers!”) may activate social conversational norms [81], which could help to personify an AI chatbot and allow consumers to treat it as a social interaction partner to converse with [82].

Lastly, we found that consumers exhibit greater preferences for and a willingness to self-disclose to search engines over AI chatbots due to greater familiarity. While AI chatbots may have been adopted by consumers, instead of abruptly switching to a different online platform, retail firms may wish to adapt their strategies to better understand consumer preferences and behaviors first by collecting data about consumer responses to search engines. The additional integration of AI Overviews on Google may allow retail companies to provide a more targeted advertising experience based on consumer interactions with AI and to reassess their search engine optimization (SEO).

5.3. Limitations

This research focuses on the effects of different information sources—namely, AI chatbots and search engines—on consumer search preferences and self-disclosure intentions, and how these effects are moderated by relationship quality. However, this research has limitations that provide valuable opportunities for future research.

First, this research is focused on young adults, such as those belonging to Gen Z, as they are more likely to adopt new technologies early as “digital natives” and potentially serve as beta testers for AI services. While students are typically used in academic experimental research, our findings could also be explored in other segments, such as older adults, as they might respond to communication cues differently from younger adults [83]. In addition, our student population sample may represent particular social classes and tended to skew female; this homogeneity may reduce the variability in responses. However, Druckman and Kam [84] have suggested that there is no difference between student and non-student samples in terms of a variety of factors, such that selecting any one issue may represent mundane realism as opposed to any real effect.

Second, the controlled environment of the experiment may limit generalizability as our experiment asked respondents to examine a static search engine results page versus an AI chatbot-generated page that did not allow for interaction. We designed the stimuli to closely replicate the visual format and content of real search engine results (e.g., Google) and AI chatbot outputs (e.g., ChatGPT) to achieve high visual fidelity and enable meaningful comparisons. By controlling extraneous variables and employing random assignment, we ensured that observed effects can be attributed to the manipulated independent variable, which enhances internal validity. However, users would continue to interact further with a page to seek out more information and spend more time on the process. This trade-off between internal and external validity may limit real-world generalizability. Thus, to enhance ecological validity, future research could examine these findings in real consumer environments—for example, using field experiments or longitudinal study.

Measurement issues regarding familiarity and self-disclosure warrant consideration. Familiarity is a clear and unidimensional construct [42,85]. Prior research has demonstrated that single-item measures are appropriate when the construct is unambiguous, unidimensional, and clearly defined [86,87,88]. 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]. Regarding self-disclosure, our research specifically focuses on the depth of self-disclosure—consumers’ willingness to share personal or sensitive information (e.g., financial data) rather than superficial information sharing (e.g., book preferences). 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]. Since consumers’ privacy concerns online, this narrow conceptualization is meaningful for capturing behavioral intent in our context.

That said, multiple-item measures can strengthen internal consistency and construct validity. While our research reflects the conceptual and contextualized use of single-item measures, it leaves room for methodological enhancement for future investigation. Thus, future research should extend this work by employing multi-item measure to capture broader aspects of familiarity and self-disclosure.

5.4. Recommended Future Research

The introduction of AI chatbots in November of 2022 set off a race in which US-based services (including Open AI’s ChatGpt, Anthropic’s Claude, and Google’s Gemini) began competing for users for their paid subscription models and offering a pared-down product for free to the broader market. Additionally, new services continue to become available, including France’s Mistral and China’s Deepseek, providing more options to users. Each offers consumers the opportunity to seek out and find products and services by entering search terms and iterating the results. However, little research has examined the variety of factors involved with information search and how they interact with the use of AI chatbots. Researchers could study differences in the search functions of these services and consumer responses to them. Our research provides one avenue to examine these search methods—namely that of familiarity and relationship quality—but delving more deeply into the product choice process, the initial and follow-up queries and prompts, the types of results, and the value of these findings to consumers and how they affect their likelihood of purchase would be illuminating for marketers as they plan product and promotional strategies.

As the adoption of AI chatbots by consumers continues to grow and evolve, consumers may build more experience with and develop knowledge and skills regarding AI chatbots search, which could alter their attitudes toward these tools. Although our research provides a useful conceptual framework to understand the future adoption of AI chatbots for searching, future research could also examine consumer perceptions of AI chatbots compared with search engines, such as familiarity, fairness, or bias perceptions, via longitudinal studies to capture any changes over time. Further, while our research focused on consumer information search for a physical product, future research could explore information search for services. Services are more complicated and intangible, and more research should examine the intricacies of searching for services such as trips. This is the case for AI chatbots, as demonstrated by MIT Technology Review’s article that explains how to use an AI chatbot for vacation searches [91].

Researchers could also consider cross-cultural examinations of AI chatbot use in product and service search. ChatGPT is available in 188 countries, and consumers across the world may be using the service for product search. For example, Park and Jun [92] found differences in perceived risk in online shopping behavior among cross-cultural samples. Similarly, Pavlou and Chai [93] found differences in online shopping behaviors between US and Chinese consumers based on Hofstede’s dimensions. Therefore, examining cross-cultural differences in the use of AI chatbots would provide insights into different potential target markets.

Lastly, future research would examine shopping behaviors—both in-store and online—using AI chatbots would elucidate consumers’ use of technology and the methods they employ to locate, sample, decide upon, and purchase a variety of products and services. Retailers may be interested in developing generative shopping bots to direct consumers to particular offerings or to provide convenience or additional value.

In conclusion, this research offers a useful perspective to better understand consumers’ responses to and evaluations of the use of AI chatbots in comparison to traditional search engines. We hope that the current research opens new doors to understanding how search engine technology can help explain consumer psychology and search behaviors in a rapidly changing society.