Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Humanities and Social Sciences Communications volume 13, Article number: 1222 (2026)
1630
Metrics details
Artificial intelligence (AI)-supported chatbots have become a common communication interface between organizations and their stakeholders. Although these systems are designed to enhance efficiency and accessibility, failures in chatbot interactions may generate unexpected communication tensions at the user level. This study examines how chatbot interaction failures may evolve into micro-level communication crises from the user experience perspective. Using an Interpretative Phenomenological Analysis (IPA) approach, in-depth interviews were conducted with individuals who had recently interacted with organizational chatbots. The findings reveal that chatbot failures are not merely technical malfunctions but may function as communicative ruptures that weaken trust, generate frustration, and potentially trigger reputational risks for organizations. The study also identifies cognitive inclusivity—the ability of AI systems to recognize the diverse emotional and informational needs of users—as a preventive mechanism that may reduce the escalation of interactional failures. These findings contribute to the crisis communication literature by conceptualizing chatbot interaction failures as micro-level communicative dynamics capable of escalating into broader corporate communication crises.
In the literature, a crisis is defined as a major event resulting in negative outcomes (Fearn-Banks and Kawamoto, 2024) that threatens an organization’s performance and stakeholder expectations (Coombs, 2007), as well as its existing structure, routine operations, or survival (Faulkner, 2001, p. 138). The fundamental perspective regarding crises focuses on “when” they will occur rather than “whether” they will occur (Coombs, 2007).
The continuous evolution of social media and information and communication technologies (ICT) has rendered the crisis ecosystem increasingly complex (Coombs and Holladay, 2012; Lee, 2020). This situation further heightens the significance of the problem of when a crisis will occur. Indeed, the dissemination of a shocking mobile phone video showing a security officer violently removing a passenger from an overbooked United Airlines flight (Victor and Stevens, 2017) and remarks by Mike Jeffries, then-CEO of Abercrombie and Fitch, stating in an interview that he wanted only “cool“ and “popular” kids to wear the brand’s clothes (The New York Times, 2014), have triggered crises. Consequently, there exists a reality where a single word, post, or image shared in digital environments can confront businesses with major crises (Fichet et al., 2016).
While the integration of businesses into digital developments increases the speed of the emergence and growth potential of crises, it also makes paracrises more visible to stakeholders (Coombs and Holladay, 2012). Paracrises are crisis-like events that have the potential to escalate into full-fledged crises if not managed at an early stage. Therefore, environmental scanning is critical in crisis management (Lee, 2020). In this context, it is crucial to examine the mechanisms, motivations, timing, and objectives underlying the adoption of digital technologies by businesses.
AI-based applications are one of the key tools in the context of digital technologies. Just as businesses previously adapted to network technologies, they are now actively incorporating artificial intelligence technologies into their business processes. Indeed, AI-based applications are becoming one of the fundamental tools in organizations’ communication with stakeholders (Florea and Croitoru, 2025). To date, numerous artificial intelligence features have been utilized in sales and advertising channels. Chatbots are one of the most widely used AI-based technologies by businesses.
Bots, a term derived from “robot” in informatics, are programs designed to execute a specific assigned task. According to another definition, a bot is a machine that automatically performs tasks controlled by a computer (Ferrara et al., 2016; Yamanoue, 2017; Zepeda, 2012). In recent years, with the development of AI-based, communication-oriented chatbots, these technologies have begun to become prevalent in corporate communication processes.
Chatbots are employed by businesses to establish social interactions with stakeholders, provide customer service, and enhance customer satisfaction and engagement (Maniou & Veglis, 2020). AI-based systems possessing deep learning and content generation capabilities offer public relations professionals not only a tool to derive insights from massive data but also a system to autonomously respond to tweets, questions, complaints, posts, and other messages on social media (Panda et al., 2019). With recent advancements, these systems have evolved beyond merely generating autonomous responses to a structure that interacts directly with the user. In this respect, chatbots now function as automated virtual assistants that automatically generate content, interact with humans on social media, attempt to mimic and potentially alter human behavior (Ferrara et al., 2016; Galitsky, 2019), provide problem-solving information to relevant stakeholders, and engage in meaningful conversations with users (Men et al., 2023).
Given the popularity and widespread use of artificial intelligence, professional communicators and researchers must be prepared for AI-based communication crises. Although AI applications offer innovative and exciting results, they also present unique and complex challenges for corporate communication professionals. Indeed, in AI-based communication crises, it is often impossible to identify a clear victim or responsible party, as is the case with traditional crisis communication strategies (Prahl and Goh, 2021). The fundamental reason why communicators need to be prepared for such crises is that artificial intelligence applications often do not go as planned and can even result in disaster. According to a Gartner (2019) report, one in four organizations using artificial intelligence experiences failure in nearly half of their AI projects due to factors such as skill gaps and unrealistic expectations. This situation is concerning for communicators, especially considering that artificial intelligence is entering high-risk sectors such as transportation, healthcare, and public safety (Prahl and Goh, 2021). In this context, organizations must be prepared for any type of crisis (Lee, 2020). Therefore, it is crucial to identify signs of potential communication crises, collect and interpret data, and respond accordingly during the pre-crisis period (Mikušová and Horváthová, 2019; Qadir et al., 2016). Otherwise, if preventive measures are not taken against crises, organizations will inevitably face crises (Taneja et al., 2014).
The literature shows that studies on the use of artificial intelligence-based chatbots in businesses largely support an optimistic perspective (De Cicco et al., 2020; Lo Presti et al., 2021; Persons et al., 2021; Sheehan et al., 2020; Trivedi, 2019; Zhang et al., 2022). However, for this optimistic approach to be valid in terms of the communication process, it depends on the communication established with the user being error-free and effective. If AI systems fail to understand the user correctly or establish interaction, these tools, which are expected to be proactive sources of information, may potentially trigger a communication crisis by exacerbating the user’s problems.
In this context, despite the increasing popularity of artificial intelligence, professional communicators and researchers must anticipate and manage the potential crises that these technologies may cause. Studies in the literature show that the use of chatbots in businesses is predominantly focused on consumer behavior (Araujo, 2018; Lo Presti et al., 2021; Pantano and Pizzi, 2020; Trivedi, 2019; Yoon et al., 2013), service quality (Belhadi et al., 2023; Donthu et al., 2021; Lin et al., 2022; Rahman et al., 2023; Zhu et al., 2023), and customer relationship management (Ashfaq et al., 2025; Kim et al., 2023).
Although several studies have examined chatbots based on user experiences (Brandtzaeg and Følstad, 2017; Hill et al., 2015; Jain et al., 2018; Kim et al., 2023; Schuetzler et al., 2020), research focusing on crisis situations is limited. These studies focus on the impact of chatbots on stakeholder management in crisis communication (Xiao and Yu, 2025), their use in crises arising during emergencies (Hofeditz et al., 2019), and perceptions regarding the use of chatbots for news access during crisis processes (Maniou and Veglis, 2020). However, none of these studies have addressed chatbots in the context of potential communication crises during the pre-crisis period. Consequently, this research aims to analyze the fundamental factors that could lead to potential communication crises in chatbot usage, as well as preventive mechanisms, based on real user experiences.
Despite the growing use of AI-powered chatbots in organizational communication, most existing studies have focused on technological efficiency, usability, or service automation. Comparatively less attention has been paid to how users experience communication breakdowns during chatbot interactions and how these experiences may evolve into communication crises at the micro-level. In particular, the literature lacks a phenomenological perspective on how users emotionally, cognitively, and relationally interpret chatbot failures. Addressing this gap is important because communication crises often emerge not only from large-scale events but also from micro-level communicative ruptures experienced by individuals in digital interaction environments.
Crisis communication is broadly defined as the collection, processing, and management of information regarding crises. The crux of crisis communication lies in how an organization interacts with its target audience and the actions it takes during various phases of a crisis (Coombs, 2012). Consequently, the field focuses on the production of effective messages by organizations during crises and the dissemination of these messages through appropriate channels (Ki and Nekmat, 2014; Ozanne et al., 2020). This focus is vital because crises can significantly alter public perceptions of trust, reputation, and loyalty toward organizations (Barton, 2001; Christensen and Lægreid, 2020). Consequently, organizations may face crises not only with their customers but also with broader stakeholder groups (Telang & Deshpande, 2016). Therefore, communication strategies must identify the priorities valued by stakeholders and develop approaches accordingly (Fombrun and van Riel, 2003). Communication has always been a valuable tool in crisis management (Coombs and Holladay, 1996).
Globalization and advancements in communication technologies have shifted communication habits and their effects on society. Rapid and widespread communication practices, particularly those occurring via social media, have triggered potential communication crises and created new challenges in crisis management (Martín-Herrera and Belda, 2021). This landscape necessitates that organizations possess dynamic capabilities, including the capacity to sense crises, seize new opportunities, and reconfigure resources (Guo et al., 2020).
In the process of resource reconfiguration (Guo et al., 2020), businesses are investing heavily in artificial intelligence (AI) technologies and chatbots with the promise of reducing costs, increasing efficiency, and providing uninterrupted services. These technologies have generated significant expectations in the business world because of their potential to automate customer service and effectively meet consumer demands (Cheng et al., 2022; Daugherty et al., 2019). However, this technological optimism does not always align with reality in the field. The misalignment between expectations and reality poses serious risks to organizations.
At this juncture, it is essential to distinguish between technical and phenomenological failures. While technical failure refers to a functional glitch or software error, phenomenological failure occurs when the user experiences a breakdown in meaning, trust, and communicative flow. The assumption that chatbots will fully replace human interaction often exposes businesses to unexpected service failures (Prahl and Goh, 2021). Although chatbots possess technically advanced natural language processing capabilities, they harbor structural deficiencies in contextualization and empathy, which are the cornerstones of human communication.
From a theoretical perspective, it is possible to address the problems arising from chatbot interactions at three different analytical levels. The first level, service failure, covers functional and operational errors, such as the system being unable to find the bill or providing incorrect directions. The primary outcome of this situation is user dissatisfaction. The second level, user disappointment, is an emotional response arising from not receiving the expected speed or solution, which raises doubts about the organization’s competence. However, the truly critical distinction is the communication crisis at the micro-interaction level of the organization. At this stage, the error ceases to be a technical flaw and becomes an ontological disconnect due to the system’s lack of empathy, inability to grasp context, and refusal to view the user as a “communication subject”. The user’s characterization of the chatbot as a “talking wall” indicates that the interaction has evolved into a reputational crisis that questions the value the organization places on the user, rather than being merely an operational failure. Therefore, this study defines chatbot errors not merely as technical “bugs”, but as early stage crisis signals that begin at the micro-interaction level and fundamentally undermine institutional trust.
Lin et al. (2022) emphasize that “disaffordances” such as limited understanding and a lack of emotion in chatbots create significant discomfort on the user’s side. Similarly, Cheng et al. (2022) stated that consumers often perceive chatbots as rigid tools devoid of empathy that provide standardized responses, a situation that complicates trust-building. These tools, which are unable to capture the nuances of human communication, cause a deep communicative disconnect between the user and the organization, especially in complex or emotionally charged situations.
Such interactional failures can escalate beyond a simple service error to a corporate reputation risk. Prahl and Goh (2021) utilized the concept of “Rogue Machines” to explain how AI can lead to an immediate communication crisis by acting out of control or providing responses disconnected from the context. As seen in the example of Microsoft’s Tay, AI can unexpectedly generate offensive or inappropriate content, causing public relations disasters.
From the user’s perspective, for an individual under stress or expecting an urgent solution, a simple technical glitch, combined with the feeling of not finding an interlocutor quickly, evolves into a crisis. In such cases, the fact that AI is an agent with ambiguous causality and intentionality further complicates the crisis management. This highlights the critical importance of an organization’s crisis-sensing ability (Guo et al., 2020). Therefore, sensing signs of potential communication crises, collecting data, interpreting it, and responding during the pre-crisis period are vital (Mikušová and Horváthová, 2019; Qadir et al., 2016).
Consequently, a deep paradox and communicative gap emerge in both literature and practice: while businesses position chatbots as proactive data collection and efficiency tools, users position these tools as corporate interlocutors for their problems. This divergence in approach between businesses and users paves the way for potential communication crises. This necessitates an understanding of the factors triggering potential communication crises in AI-supported chatbot interactions. This necessity is directly linked to the dynamic capability of “sensing the crisis and seizing new opportunities” (Guo et al., 2020). Based on this premise, this study focuses on the following main research questions:
RQ1: In AI-powered chatbot interactions, what are the key triggers that transform a technical communication process into a potential communication crisis, and what are the breaking points in the user experience?
RQ2: In the context of user experience, what preventive mechanisms can prevent the escalation of a potential communication crisis and restore trust?
Taken together, these studies suggest that chatbot failures should not be evaluated solely based on technical malfunctions. Instead, they represent interactional breakdowns at the intersection of technology, emotion, and organizational communication. However, the existing literature rarely connects chatbot interaction failures with crisis communication dynamics, particularly from the perspective of lived user experiences.
This study aimed to understand the context of participants’ experiences in conversations in which they engaged with chatbots as corporate interlocutors. Accordingly, we conducted an in-depth examination of the factors affecting satisfaction, process functioning, quality, and problem-solving performance during chatbot usage. The primary objective of this research was to identify—based on these experiences—the main factors that could trigger potential crises in future processes and the preventive mechanisms that could mitigate them.
To obtain deeper and richer insights from the interviews (Miller and Barrio Minton, 2016; Padilla-Diaz, 2015; Pietkiewicz and Smith, 2014), we employed Interpretative Phenomenological Analysis (IPA), a qualitative method. Small sample sizes are typical in IPA studies to allow for detailed and nuanced analyses (Bartholomew et al., 2021; Smith and Osborn, 2003). The literature suggests that while the number of participants may vary between one and thirty, depending on the variables, there is a tendency toward the lower limit of one. As the sample size decreases, the focus on a specific phenomenon increases, enabling a more in-depth analysis of the results (Brocki and Wearden, 2006; Eatough and Smith, 2017). In this study, the quality and depth of the data were prioritized over their representativeness (Smith, 2004).
We collected data between May and June of 2025. Prior to the study, ethical approval was obtained from the Ethics Committee of Tokat Gaziosmanpaşa Üniversitesi (Date: 29.05.2025, Decision No: 01-121). The research process was conducted in accordance with the Declaration of Helsinki, and informed consent was obtained from all the participants.
The interview questions were collaboratively developed by an academic team with experience in the subject matter and expertise in qualitative research. We utilized a purposive sampling technique to identify participants with the potential to offer appropriate, rich, relevant, and diverse information regarding the research purpose (Creswell and Creswell, 2022; DiCicco-Bloom and Crabtree, 2006). Subsequently, snowball sampling was used to expand the sample.
The inclusion criteria were as follows:
Having communicated with an organization via a chatbot during product, service, or communication processes,
Having had a recent chatbot experience,
Residing in Türkiye.
To increase sample diversity, we shared research information via social media platforms to reach individuals who met these criteria. We conducted semi-structured interviews with 12 participants. This method is preferred because it elicits rich, detailed, and inclusive first-person accounts of lived experiences (Polkinghorne, 2005). Because the participants were in different geographical locations, interviews were conducted via online platforms (Zoom) or submitted via digital forms based on participant preference. The voice interviews lasted approximately 15–30 min, were recorded, and subsequently transcribed. The data collection process was concluded when participants began to repeat similar information, indicating that data saturation had been reached.
The participant group consisted of seven men and five women with varying professions, ages, and IT skills (Table 1).
Data analysis was performed in five basic steps using an inductive approach. Merriam and Tisdell (2016) state that inductive analysis allows researchers to combine, reduce, and interpret data. We utilized MAXQDA software to manage the qualitative datasets (MAXQDA, 2020). All open coding, data reduction, and theme development processes were conducted using this program.
The analysis process consisted of the following stages:
Data verification: Transcripts generated by AI-supported tools were listened to simultaneously with the audio recordings to verify accuracy and correct any errors.
Open coding: Each word, phrase, and paragraph related to the research questions was coded, resulting in 154 distinct codes.
Categorization: The supercode/sub-code structure of the codes was checked and reclassified.
Theme development: Using the “Creative Coding” feature of MAXQDA, repetitive codes were grouped together. Consequently, two main categories were created: “Experiences” and “Demographics.” The Experiences category contained 106 codes under two super-codes and five sub-codes, while the Demographics category included 48 codes under four super-codes (Table 1A – Appendix). The coding process was conducted simultaneously by two independent researchers. The rigor of the Interpretative Phenomenological Analysis (IPA) was maintained through a collaborative coding process involving two independent researchers. To ensure structural consistency, a detailed codebook was developed that explicitly defined the main and sub-themes along with prototypical sentence examples and coding patterns. The intercoder agreement rate was 0.97 (calculated using MAXQDA). This figure represents the consistency in the structural categorization of the data, rather than a mechanical consensus on its final meaning. In nine instances where coding disagreed, the researchers held deliberative sessions to reach a consensus, ensuring that reflexive interpretation remained the core of the analysis while achieving intersubjective stability. Calculated using MAXQDA software, the inter-coder agreement rate derived from the two separate coding sets was 0.97. Despite the fact that the coders produced 314 instances of overlapping coding, they also produced nine distinct coding instances. The nine discordant codings that impeded the attainment of consensus were deliberated and resolved by the two coders. Consequently, it can be argued that this research achieved a comprehensive consensus.
Interpretation and reporting: In the final stage, themes were discussed in briefing sessions with the research team to ensure the representation of the participants’ experiences and to relate the findings to the literature. Based on the coding, general themes containing the dynamics triggering potential crises and the buffering mechanisms preventing them were determined (Table 2A – Appendix).
All 154 codes in the dataset were included in the analysis; participants’ experiences were interpreted along the axis of “crisis trigger” (negative) and “crisis preventer” (positive) mechanisms. Within this scope, three main themes were identified (Table 2A-appendix). The findings are presented under each main theme heading.
Technology-induced anxiety encompasses negative emotions, apprehension, and fear that emerge following the use of technology (Kummer et al., 2017). In the communication process with AI-based technologies, the element of urgency acts as a factor influences AI failures. As urgency intensifies, the importance attributed to AI’s empathy responses diminishes (Guo et al., 2025). The analyses reveal that users do not initiate chatbot interactions with neutral motivation; on the contrary, they experience the process through a “forced orientation” driven by existing grievances and high expectations. The presence of “operational service disruptions” (n = 5) and “digital purchasing process issues” (n = 7) in the code distribution indicates that users approach the system not out of technological curiosity but out of a necessity to find urgent solutions to their problems.
This scenario highlighted a critical dynamic in which users’ expectations for technological proficiency were at their peak, whereas their tolerance for errors was minimal. As participants viewed the chatbot as a last resort for problem resolution, any technological inadequacy encountered was perceived as far more than just a technical error. Instead, it serves as a catalyst for profound disappointment and a trigger for potential crises. Participant 12’s statement illustrates the tension between the technological capacity promised by the system (i.e., high resolution rate) and the user’s actual expectations:
“At first, I did not actually realize that I was interacting with a chatbot. I was having issues with a product I purchased, and technical support was not providing a solution… It was only when I entered my complaint into the system that I realized I was interacting with a chatbot.” (P12)
In a similar manner, Participant 10 was compelled to regard the chatbot as a timely savior in an uncertain environment outside working hours. However, this state of “necessity” was a factor that increased the severity of the potential crisis that would arise if the technology proved inadequate.
“I could not get an invoice for a vacation I took. I was working very hard, and it kept coming to my mind at night. However, customer service representatives did not work after 12:00. While researching on the tourism company’s website, I came across a chatbot.” (P10)
This user tendency was observed to develop in instances where human-based channels fell short of users’ expectations, often due to factors such as disinterest or rejection. Participant 5’s experience demonstrated that dissatisfaction with conventional channels engendered an expectation of compensation from the chatbot and that the failure to meet this expectation posed a risk.
“After making a purchase, I had problems with the product that I received. I contacted customer service, but they did not care. So I used a public website to… complain.” (P5)
In this context of low tolerance, the most powerful mechanism capable of overcoming the “perception of technological inadequacy” and preventing a potential crisis was “emergency response speed and success capability.” “Operational speed and efficiency” (n = 8) and “Results-oriented success” (n = 7), found in positive codes, proved technological adequacy and mitigated crisis risk by immediately responding to the user’s high expectations. Participant 5 described how their expectations were met because of the system’s analytical capability (technological competence) as follows:
“First, it was very fast and solution-oriented. It analyzed the text I wrote, asked me questions, and gathered detailed information on the subject. It was very easy to use, and I didn’t need to contact customer service.” (P5)
In this context, the concepts of speed and solution-orientedness were not merely a matter of convenience; they were strategic factors that mitigated the user’s suspicion of technological inadequacy in the system. Participant 1, despite their stated desire for human contact, asserted that the system’s rapid problem-solving capabilities (its ability to meet expectations) served as a compensatory mechanism for this deficiency. They emphasized that the management of crisis potential is contingent upon high technological performance.
“I must admit that even though my problem was solved quickly, the feeling of interacting with people is different. Of course, I was happy that the problem I experienced was solved quickly.” (P1)
Participants’ characterization of their chatbot experiences using metaphors such as “unresponsive interaction” and “mechanical monologue” is analyzed in this study through the lens of the “Lack of Empathy Response” concept.
The empathy capacity of artificial intelligence refers to the ability to perceive, understand, and respond to human thoughts, emotions, behaviors, and experiences, representing the embodiment of AI’s social nature of AI (Murphy et al., 2019). Research on AI-based services has predominantly focused on empathy. Interactions with AI tools possessing high empathy responses have been shown to reduce perceived anxiety and frustration (Brave et al., 2005; Mehmood et al., 2024) and enhance perceptions of communication quality (Hone, 2006; Chi and Vu, 2023). An empathy response that resolves problems and improves experiences increases user communication continuity (Lv et al., 2022). However, the transformation of user-chatbot interactions into a “mechanical monologue” reinforces the user’s sense of being disregarded, leading to affective detachment. When the user ceases to be an active agent trying to solve a problem and becomes passive in the face of an unresponsive interface, the experience transcends a simple technical glitch, creating a profound communicative break.
The interaction process between the user and AI was the most critical finding of this study. Participants’ negative experiences describe a process of “communicative alienation” (n = 2), where they are ignored as communicative subjects rather than merely encountering technical errors (Table 1A). Participants did not experience this process as a reciprocal dialogue but as a mechanical data query process that excluded the human element. This sentiment is clearly reflected in Participant 12’s statement.
“In the chatbot application, I felt as if I were performing a search via a search engine on the company’s website rather than actually chatting.” (P12)
When combined with the chatbot’s inability to grasp context and its constant repetition, this sense of alienation gives rise to the “unresponsive interaction” phenomenon that we have discussed. The participants’ feeling of speaking to a wall indicates that the system fails to provide reciprocity, leaving the user in a communicative void. The AI’s failure to analyze questions in depth due to “cognitive insufficiency” (n = 6) traps the user in a helpless loop (Table 1A). Participant 2 described this vicious cycle and the feeling of being misunderstood:
“The chatbot started by writing, ‘Welcome, how can I help you?’ When I wrote down my problem, it said, ‘Please provide more information.’ However, no further information was provided. We could not establish mutual communication. I tried to explain my trouble in my own way, but the chatbot kept trying to sustain the conversation with the same phrases.” (P2)
Similarly, the system’s failure to perceive even basic human courtesies and its experience of “semantic discord” (n = 1) completely severed communication (Table 1A). Participant 7’s experience reveals how disconnected the AI is from the context and the astonishment this creates for the user:
“It wrote an answer to my question… I said, ‘Thank you.’ Instead of saying ‘You’re welcome,’ the chatbot copied and pasted the exact previous answer (the service hours warning).” (P7)
This situation demonstrates that chatbots are plagued by “interactional sterility” (n = 4). Unable to go beyond pre-scripted answers and scenarios, the system transforms the dialogue into a mechanical monologue (Table 1A). Participant 6 interpreted the chatbot’s presentation of only pre-loaded responses as a lack of quality and criticized the system’s inability to deepen the interaction:
“I think the fact that it simply presents questions loaded by the shopping site and their corresponding answers shows how low the quality of the chatbot is… I think it needs to understand me and deepen what I have written.” (P6)
Consequently, when users cannot find an interlocutor who understands them, they experience a profound sense of empathy deprivation regarding the system and brand. Participant 11’s statement summarizes the frustration created by this process and the rejection of robotic communication:
‘I think I encountered a robot that just gives generic [or ordinary] answers. I wrote ‘Thank you’ to the chatbot, but it kept copying the previous answers. ‘ (P11)”
User experiences revealed that chatbots must possess a “cognitive and social inclusion” mechanism to serve as an antidote to communication gaps and potential crises. The codes “semantic alignment and comprehensibility” (n = 4) and “sense of social presence” (n = 5) in the data set proved that the user’s satisfaction level increased when they felt “heard,” and that this acted as a crisis-preventing buffer (Table 1A).
The AI’s accurate understanding of the user’s intent (cognitive depth) made them the subject of communication again and prevented alienation. Participant 1’s experience was the clearest example of how the problem evolved into a solution when the chatbot deepened the issue by achieving semantic alignment:
“But as I started talking to the chatbot, I was very surprised. I felt like it understood me, and I really felt like there was someone there. It asked me questions to help me elaborate on my complaints… The chatbot asked, ‘Has it been 24 h since your order’s delivery time?’ When I said, ‘No,’ it said, ‘That’s why you could not find it. Please wait 24 h after your order.’ By following the chatbot’s instructions, I was able to resolve the issue I was experiencing.” (P1)
Similarly, Participant 5 emphasized that the chatbot’s analytical ability (comprehensiveness) established trust without the need for human interaction:
“It analyzed the text I wrote and asked me questions to gather detailed information about the subject. It was very easy to use, and I didn’t need to contact customer service.” (P5)
In addition to cognitive depth, chatbots’ ability to create a “sense of social presence has emerged as a crisis-preventing mechanism. The statements of Participants 3 and 10 showed that when mechanical coldness gave way to a perception of human interaction, the users’ trust in the system and their positive surprise increased:
“When I wrote my problem to the chatbot, it asked me some probing questions. This was surprising. I felt as if someone was talking to me.” (P3)
“First, it tried to analyze my problem in detail, he said. At one point, I thought I was talking to a real person.” (P10)
Research on human-robot interaction (HRI) and human-computer interaction (HCI) suggests that trust in human-AI relationships may parallel human-human trust (Lewandowsky et al., 2000; Schniter et al., 2020). Gulati et al. (2018), in their study conducted via Siri, demonstrated that trust is influenced by perceived helpfulness (i.e., reputation for benevolence), reliability (i.e., consistent and recent feedback), and the ability to meet the participant’s needs (i.e., competence). However, the increasing reliance on AI introduces challenges related to consumer trust and privacy. Issues such as a lack of algorithmic transparency, potential data misuse, and diminished consumer autonomy can undermine perceptions of reliability and integrity in AI systems (Daugherty et al., 2019), thereby fostering a sense of “digital vulnerability” among users. One potential psychological consequence of AI-human interaction is that the opacity of the algorithmic structure raises concerns regarding consumer autonomy and security (Pasquale, 2015).
The most fundamental finding obtained in this study is that a failed interaction transforms into “perception of trust and privacy violation” (n = 2) and “loss of time and efficiency” (n = 6), indicating that a technical issue can evolve into a potential communication crisis (Table 1A).
Technology that promised speed and convenience but resulted in time loss created a feeling of “deception” in the user. More critically, the necessity of entrusting personal data to a system that lacked empathy and could not resolve the issue (functional deadlock, n = 2) was perceived as a threat by the user (Table 1A). This stage was the breaking point at which service failure turned into an organizational crisis. Participant 5’s doubts about data security clearly revealed the extent of this perceived threat:
“But there is another situation here that bothers me. Data were collected through the chatbot. It is unclear how and where these data are used. This could lead to a determination of my private life. That is why I think it is important to be careful.” (P5).
Another factor contributing to the erosion of trust was that the promised speed had led to a vicious cycle. Users interpreted constantly encountering the same responses while waiting for a solution as a waste of time. Statements from Participants 2, 8, and 12 summarized the impatience created by this process:
“I think the quality was very low. I’m having a problem. Instead of being solution-oriented and trying to understand me, the chatbot wasted my time with the same phrases over and over.” (P2)
“It was a complete waste of time. I was dissatisfied because it wasn’t solution-oriented.” (P8)
“I don’t want to waste time again with a chatbot that copies and sends the same answer several times.” (P12)
At this stage, it was argued that alongside improvements in AI quality, the emerging “hybrid communication capability (human routing)” acted as the strongest safety valve, preventing the emergence of potential crises. Offering the option to consult a human when technology reaches its limits, or AI assumes an educational role, eliminates the user’s feeling of helplessness and perceived threat, thereby restoring trust in the brand. Participant 3 expressed how the system’s hybrid and educational structure changed their perception of quality:
“The hybrid design of the system reveals its quality. In addition, I was impressed that it analyzed what I wrote beyond simple answers and asked new questions to detail the situation. Moreover, its educational structure that taught me things also increases its quality, in my opinion.” (P3)
The ability to provide fast and results-oriented solutions with hybrid communication capabilities directly increased “general user satisfaction” (n = 14) (Table 1A). The experiences of Participants 1 and 9 demonstrated how, when the system worked correctly, a crisis could be turned around and transformed into satisfaction.
“Using the chatbot was very simple. Therefore, I was able to get my work done without getting caught up in many tedious processes, such as introducing myself to customer service. I didn’t wait in line. I do not have to explain anything to anyone. The system, which I could easily use, produced a solution to my problem.” (P1)
“Yes, I would like to use it again. Because it was fast and results-oriented.” (P9)
Otherwise, when these trust mechanisms (hybrid structure, speed, solution) did not come into play, “general user dissatisfaction” (n = 13) became inevitable (Table 1A). This situation could trigger a potential crisis. The statements of Participants 6 and 8 clearly show that users were not closed to the system but would not tolerate it when faced with “insufficient infrastructure”:
“I was not satisfied with this experience. However, I do not believe that this means that these chatbots are bad or inadequate. I would like to experience a chatbot designed with the right artificial intelligence infrastructure in a crisis situation.” (P6)
“I do not want to use them. This is because they are inadequate. In a crisis situation, when a problem occurs, people are stressed. We should not waste people’s time during this process. For this, I think there needs to be a serious artificial intelligence infrastructure.” (P8)
In this study, cognitive inclusivity refers to the capacity of AI communication systems to recognize and respond to users’ diverse emotional, informational, and contextual needs. Systems lacking cognitive inclusivity may unintentionally escalate communication tension during interaction failures.
This study analytically distinguishes between triggers and preventive mechanisms. Triggers refer to the factors that initiate communicative ruptures during chatbot interactions, such as misunderstanding the user’s intent, repetitive responses, or lack of contextual sensitivity. Preventive mechanisms, on the other hand, refer to system capacities that mitigate or prevent escalation of these failures. In this sense, triggers operate at the interactional level, whereas preventive mechanisms function at the design and system levels. This analytical distinction allows us to separate the causes of communicative breakdown from the institutional capacities that may prevent their escalation into communication crises.
This study examines the role of AI-powered chatbots in user experience, not merely as a technological performance metric but as a potential crisis dynamic that tests the trust relationship between the organization and the user. Numerous studies exist in the literature on chatbots have been conducted. While a significant portion of these studies support an optimistic perspective on chatbot usage (Balderas et al., 2023; De Cicco et al., 2020; Maniou and Veglis, 2020; Persons et al., 2021; Zhang et al., 2022), this study reveals that chatbots also represent a communication risk area for organizations.
Among the fundamental characteristics of chatbots is their ability to provide consistent and standardized responses (Chow et al., 2023; Yu et al., 2024). However, during a problem or crisis, chatbots are expected to generate flexible, personalized, and context-specific responses (Xiao and Yu, 2025). Yoon and Kang (2026) also emphasize the need to harmonize the expected relationship dynamics in chatbots. In this context, the fundamental expectation is the elimination of emotional uncertainty and the provision of practical guidance for effective problem-solving.
Users initially have lower expectations of communication quality than real customer representatives (Zhou et al., 2023). As the perceived level of chatbot competence increases, user confidence also increases (Cheng et al., 2022). Findings from this study reveal that users experience chatbots as a corporate contact to find solutions to their problems, and their basic expectation is that their problems will be solved. In communication that begins on fragile and stressful ground, users’ tolerance levels are quite low. Regardless of the nature of the crisis, the ability of chatbots to competently address stakeholder demands is a determinant of an organization’s competence in crisis management. This type of performance is important for improving the overall user experience and corporate reputation during challenging times for organizations (Xiao and Yu, 2025). This is because the usability and fast response time of chatbots must be considered (Ashfaq et al., 2020). The findings obtained within the scope of the research also show that the key factors that will eliminate the expectation-performance gap in chatbots are speed and result-orientation.
Yoon and Kang (2026) posited that chatbots capable of establishing deep empathy with users can positively enhance the communication process. Notably, factors such as the absence of empathy, incapacity to respond to emotional needs, and inability to establish relational bonds are underlying factors in the failure of chatbots (Griffith and Simonite, 2018; Martin, 2018; Sheehan et al., 2020; White, 2018; Yu et al., 2024). The elements that users find most disturbing when using chatbots are their limited understanding capacity and lack of emotion (Lin et al., 2022). Consumers perceive crises as negative events, which leads them to blame the actors involved (Coombs, 2007; Dean, 2004). Conversely, an enhancement in users’ perceived competence regarding chatbots has been demonstrated to engender heightened user satisfaction while simultaneously attenuating the responsibility burden on organizations during crises (Xiao and Yu, 2025).
The results reveal that chatbots’ failure to comprehend users is characterized as a “lack of empathy response.” The “crisis responsibility attribution” process outlined in Coombs’ (2007) Situational Crisis Communication Theory (SCCT) has been transferred to a digital dimension in this study; when users could not find anyone to address their concerns, they interpreted the problem not as a technical error but as an indication of the value the institution placed on them. Therefore, the findings show that chatbot failures go beyond being a simple service error; they are “crisis signals” that create ontological insecurity and threaten the brand’s reputation. Chatbots that create a sense of social presence among users strengthen emotional bonds (Araujo, 2018) and perceived trust in service interactions (Cheng et al., 2022). Research emphasizes that chatbots must create a sense of social presence (Derrick et al., 2011; Rafaeli and Noy, 2005; Zhang et al., 2012). Research results indicate that chatbots enhance user experience by possessing a “cognitive and social inclusivity” mechanism and fostering a sense of “social presence.” In this respect, the research supports existing literature.
While chatbot interaction failures may initially appear similar to routine service breakdowns, not all service failures constitute a communication crisis. In this study, a communication crisis emerges when interaction failures extend beyond functional dissatisfaction and threaten the communicative relationship between the organization and the user. In other words, while service failure may lead to temporary user frustration, a communication crisis involves deeper consequences, such as trust erosion, perceived organizational indifference, and reputational vulnerability. Therefore, chatbot failures should be interpreted as early communicative crisis signals when users begin to question the reliability, responsiveness, or legitimacy of an organization’s communication practices.
Given the unpredictable nature of crises, the success of the communication process is crucial. This necessitates the optimization of chatbot technologies in the crisis management process (Xiao and Yu, 2025). Research results also indicate that organizations need to equip their chatbot applications with cognitive inclusiveness that can correctly understand the user’s intent and hybrid capabilities that can direct them to human interaction when necessary.
As with any scientific study, certain limitations must be considered when evaluating the results of this research. The study was designed using the Interpretative Phenomenological Analysis (IPA) method and was based on the subjective experiences of 12 participants. Owing to the nature of qualitative research, the findings are not statistically generalizable; however, they provide in-depth insights.
Regarding the scope, all participants began their chatbot experience due to “an existing problem or grievance” (return, invoice, etc.). This strengthens the “crisis-focused” structure of the research, but does not cover the experiences of users who use the chatbot for different purposes. The technological infrastructure of the chatbots experienced by users (whether rule-based or Generative Artificial Intelligence/LLM-based) was not differentiated within the scope of the study. The experiential differences created by bots with different intelligence levels are beyond the scope of this paper. The research was conducted on a sample of Turkish participants. Since communicative expectations and the “desire to find an interlocutor” may vary according to cultural codes, the results may differ across cultures. Hall and Hall (1990) classified world cultures based on high and low context. In high-context cultures, communication involves indirect verbal expression and nonverbal communication, and the listener is expected to understand the meaning of the message by considering its context. In contrast, low-context communication emphasizes direct and explicit information exchange, and the listener does not have to consider complex contexts when interpreting the message. Türkiye’s cultural fabric exhibits a “high-context” structure in communication; that is, beyond the message itself, great importance is placed on the presence of the interlocutor, the tone of voice, and emotional empathy. The “interactional impasse” and the pain of “not finding an interlocutor” that emerged in this study is not only a technical inadequacy but also a conflict of this cultural expectation in the digital realm. While information-focused and direct responses may be sufficient in Western, “low-context” cultures, the expectation of users in Türkiye to find a “digital soul” or “social presence” in chatbots is a cultural reflection of these findings. Therefore, the results of this study provide a powerful model for understanding the crisis potential of chatbot-human interaction in societies with similar cultural codes (such as the Middle East, Mediterranean, and Latin America). Future studies testing the cultural boundaries of this ‘interlocutor-focused’ expectation through cross-cultural comparisons will add depth to crisis communication theory.
Based on the findings and limitations of this study, future research should compare the performance of traditional (rule-based) chatbots with next-generation chatbots powered by large language models (LLMs), such as ChatGPT, within the context of crisis management. Whether the phenomenon of “lack of empathy” diminishes in these new systems that utilize more natural language constitutes a critical research topic. Furthermore, in connection with this, concepts regarding AI technologies, such as inclusivity, alienation, and hybrid competence, can be quantified through scale development studies and tested on larger samples.
In summary, chatbot interactions are not merely functional exchanges but a dynamic process in which corporate reputation, trust, and communication expectations are continuously negotiated. The conceptual dynamics identified in this study are illustrated in Fig. 1.
Conceptual model of micro-level communication crises in chatbot interactions.
The “Artificial Intelligence-Supported Micro-Interactive Crisis Cycle” (Table 2), developed based on the data obtained from this research, presents a holistic model of how a crisis can emerge from a technical error (RQ1), at what stage it can transform into a reputation risk (paracrisis), and how it can be mitigated through “communicative repair” (RQ2). This model shifts crisis communication from static message management to proactive relationship management, focusing on human–computer interaction.
While the integration of artificial intelligence technologies into customer service processes is regarded as a strategic maneuver that leads to operational efficiency and cost advantages for businesses, this study reveals that there is considerable potential for a communication crisis on the other side of the coin. This research, which examines user experiences from a phenomenological perspective, concludes that interactions with chatbots are not merely a technical exchange of information; rather, they are a critical communication space where users’ expectations of being acknowledged, understood, and valued are tested, and where the relationship of trust between the institution and the individual becomes fragile.
The most fundamental results of this research reveal that chatbot failures are experienced in the user’s world not as a simple software bug but as a lack of empathy that leads to communicative alienation. The findings demonstrate that users do not initiate interactions from a neutral standpoint; rather, they engage in the process on a fragile and intolerant foundation shaped by their pre-existing grievances. At this sensitive stage, the cognitive shallowness exhibited by artificial intelligence is interpreted as “unresponsive interaction” potentially resulting in a communication crisis.
However, the findings of this study demonstrate that the crisis cycle is not inevitable; it can be overcome through the implementation of properly designed “preventive mechanisms”. Analyses conducted within the scope of the research demonstrate that when chatbots are equipped not only with speed-focused capabilities but also with “cognitive comprehensiveness” that understands the user’s intent and “social presence” that provides a sense of human contact, the crisis evolves into a process of “communicative repair”. In circumstances where technology has failed, the “hybrid communication capability” (human referral) has been shown to emerge as the most effective safety instrument, effectively extinguishing the user’s sense of helplessness and restoring trust.
In conclusion, this study makes a significant contribution to the existing literature on technology acceptance and efficiency by arguing that chatbots should be repositioned as strategic actors in proactive crisis management. In the process of devising strategies for chatbots, businesses must regard these tools not merely as a mechanism for reducing expenditure but as a means of providing a solution to issues by offering support to users in times of crisis and, when required, providing a human connection. This will be the key to a sustainable corporate reputation in the digital age.
The datasets created and/or analysed during the current study, and the analysis codes, are available in the repository [Repository Name: Figshare] at https://doi.org/10.6084/m9.figshare.30963347.
Araujo T (2018) Living up to the chatbot hype: the influence of anthropomorphic design cues and communicative agency framing on conversational agent and company perceptions. Comput Hum Behav 85:183–189. https://doi.org/10.1016/j.chb.2018.03.051
Article Google Scholar
Ashfaq M, Makkar M, Hoang A-P, Dang-Pham D, Do MHT, Nguyen ATV (2025) Exploring customer stickiness during “smart” experiences: a study on AI chatbot affinity in online customer services. J Res Interact Market. https://doi.org/10.1108/JRIM-09-2024-0452
Barton L (2001) Crisis in organizations II. South-Western College
Belhadi A, Kamble S, Benkhati I, Gupta S, Mangla SK (2023) Does strategic management of digital technologies influence electronic word-of-mouth (eWOM) and customer loyalty? Empirical insights from B2B platform economy. J Bus Res 156:113548. https://doi.org/10.1016/j.jbusres.2022.113548
Article Google Scholar
Brandtzaeg PB, Følstad A (2017) Why people use chatbots. In: Kompatsiaris I, Cave J, Satsiou A, Carle, G, Passani A, Kontopoulos E, Diplaris S, McMillan D (eds) Internet science. Springer International Publishing, pp 377–392
Brave S, Nass C, Hutchinson K (2005) Computers that care: investigating the effects of orientation of emotion exhibited by an embodied computer agent. Int J Hum Comput Stud 62(2):161–178. https://doi.org/10.1016/j.ijhcs.2004.11.002
Article Google Scholar
Cheng X, Zhang X, Cohen J, Mou J (2022) Human vs. AI: understanding the impact of anthropomorphism on consumer response to chatbots from the perspective of trust and relationship norms. Inf Process Manag 59(3):102940. https://doi.org/10.1016/j.ipm.2022.102940
Article Google Scholar
Chi NTK, Vu NH (2023) Investigating the customer trust in artificial intelligence: the role of anthropomorphism, empathy response, and interaction. Caai Trans Intell Technol 8(1):260–273. https://doi.org/10.1049/cit2.12133
Article Google Scholar
Christensen T, Lægreid P (2020) The coronavirus crisis-crisis communication, meaning-making, and reputation management. Int Public Manag J 23(5):713–729. https://doi.org/10.1080/10967494.2020.1812455
Article Google Scholar
Coombs WT (2007) Protecting organization reputations during a crisis: the development and application of situational crisis communication theory. Corp Reput Rev 10(3):163–176. https://doi.org/10.1057/palgrave.crr.1550049
Article Google Scholar
Coombs WT (2012) Ongoing crisis communication: planning, managing, and responding (2nd ed.). Sage Publication
Coombs WT, Holladay JS (2012) The paracrisis: the challenges created by publicly managing crisis prevention. Public Relat Rev 38(3):408–415. https://doi.org/10.1016/j.pubrev.2012.04.004
Article Google Scholar
Coombs WT, Holladay SJ (1996) Communication and attributions in a crisis: an experimental study in crisis communication. J Public Relat Res 8(4):279–295. https://doi.org/10.1207/s1532754xjprr0804_04
Article Google Scholar
Daugherty PR, Wilson HJ, Michelman P (2019) Revisiting the jobs artificial ıntelligence will create. MIT Sloan Manag Rev 60(4):1–8
Google Scholar
De Cicco R, Silva SC, Alparone FR (2020) Millennials’ attitude toward chatbots: an experimental study in a social relationship perspective. Int J Retail Distrib Manag 48(11):1213–1233. https://doi.org/10.1108/IJRDM-12-2019-0406
Article Google Scholar
Donthu N, Kumar S, Pandey N, Pandey N, Mishra A (2021) Mapping the electronic word-of-mouth (eWOM) research: a systematic review and bibliometric analysis. J Bus Res 135:758–773. https://doi.org/10.1016/j.jbusres.2021.07.015
Article Google Scholar
Faulkner B (2001) Towards a framework for tourism disaster management. Tour Manag 22(2):135–147. https://doi.org/10.1016/S0261-5177(00)00048-0
Article Google Scholar
Fearn-Banks K, Kawamoto K (2024) Crisis communications: a casebook approach. Routledge
Ferrara E, Varol O, Davis C, Menczer F, Flammini A (2016) The rise of social bots. Commun ACM 59(7):96–104
Article Google Scholar
Fichet ES, Robinson JJ, Dailey D, Starbird K (2016) Eyes on the ground: emerging practices in periscope use during crisis events. International Conference on Information Systems for Crisis Response and Management
Florea NV, Croitoru G (2025) The impact of artificial intelligence on communication dynamics and performance in organizational leadership. Adm Sci 15(2):33. https://doi.org/10.3390/admsci15020033
Article Google Scholar
Fombrun CJ, van Riel CBM (2003) Fame & Fortune: How successful companies build winning reputations. Corp Reput Rev 6(4):390–396. https://link.springer.com/content/pdf/10.1057/palgrave.crr.1540007.pdf
Google Scholar
Galitsky B (2019) A Content Management system for chatbots. In: Galitsky B (ed) Developing enterprise chatbots: learning linguistic structures. Springer International Publishing, pp 253–326 https://doi.org/10.1007/978-3-030-04299-8_9
Gartner (2019) 37% of Organizations have implemented AI in some form. https://www.gartner.com/en/newsroom/press-releases/2019-01-21-gartner-survey-shows-37-percent-of-organizations-have Accessed 15 June 2025
Gulati S, Sousa S, Lamas D (2018) Modelling trust in human-like technologies. In: Proceedings of the 9th Indian conference on human-computer interaction. pp 1–10 https://doi.org/10.1145/3297121.3297124
Guo Y, Xu L, Wang C (2025) Exploring the effect of empathic response and its boundaries in artificial intelligence service recovery. J Retail Consumer Serv 82. https://doi.org/10.1016/j.jretconser.2024.104065
Guo H, Yang Z, Huang R, Guo A (2020) The digitalization and public crisis responses of small and medium enterprises: implications from a COVID-19 survey. Front Bus Res China 14(1):19. https://doi.org/10.1186/s11782-020-00087
Article Google Scholar
Hall ET, Hall MR (1990) Understanding cultural differences: Germans, French and Americans. Intercultural Press, Yarmouth, England
Hill J, Randolph Ford W, Farreras IG (2015) Real conversations with artificial intelligence: a comparison between human–human online conversations and human–chatbot conversations. Comput Hum Behav 49:245–250. https://doi.org/10.1016/j.chb.2015.02.026
Article Google Scholar
Hofeditz L, Ehnis C, Bunker D, Brachten F, Stieglitz S (2019) Meaningful use of social bots? Possible applications in crisis communication during disasters. In: 27th European Conference on Information Systems (ECIS). https://aisel.aisnet.org/ecis2019_rp/138
Hone K (2006) Empathic agents to reduce user frustration: the effects of varying agent characteristics. Interact Comput 18(2):227–245. https://doi.org/10.1016/j.intcom.2005.05.003
Article Google Scholar
Jain M, Kumar P, Kota R, Patel SN (2018) Evaluating and informing the design of chatbots. In: DIS 2018 – Proceedings of the 2018 designing interactive systems conference. pp 895–906 https://doi.org/10.1145/3196709.3196735
Ki E-J, Nekmat E (2014) Situational crisis communication and interactivity: usage and effectiveness of Facebook for crisis management by Fortune 500 companies. Comput Hum Behav 35:140–147. https://doi.org/10.1016/j.chb.2014.02.039
Article Google Scholar
Kim J, Li Y, Choi J (2023) Understanding the continuance intention to use chatbot services. Asia Mark J 25(3):99–110. https://doi.org/10.53728/2765-6500.1613
Article Google Scholar
Kummer T, Recker J, Markus B (2017) Technology-induced anxiety: manifestations, cultural influences, and its effect on the adoption of sensor-based technology in German and Australian hospitals. InformationManagement 54:73–89. https://doi.org/10.1016/j.im.2016.04.002
Article Google Scholar
Lee SY (2020) Stealing thunder as a crisis communication strategy in the digital age. Bus Horiz 63(6):801–810. https://doi.org/10.1016/j.bushor.2020.07.006
Article Google Scholar
Lewandowsky S, Mundy M, Tan G (2000) The dynamics of trust: comparing humans to automation. J Exp Psychol Appl 6(2):104. https://doi.org/10.1037/1076-898X.6.2.104
Article PubMed CAS Google Scholar
Lin X, Shao B, Wang X (2022) Employees’ perceptions of chatbots in B2B marketing: affordances vs. disaffordances. Ind Mark Manag 101:45–56. https://doi.org/10.1016/j.indmarman.2021.11.016
Article Google Scholar
Lo Presti L, Maggiore G, Marino V (2021) The role of the chatbot on customer purchase intention: towards digital relational sales. Ital J Mark 2021(3):165–188. https://doi.org/10.1007/s43039-021-00029-6
Lv X, Yang Y, Qin D, Cao X, Xu H (2022) Artificial intelligence service recovery: the role of empathic response in hospitality customers’ continuous usage intention. Comput Hum Behav 126:106993. https://doi.org/10.1016/j.chb.2021.106993
Article Google Scholar
Maniou TA, Veglis A (2020) Employing a chatbot for news dissemination during crisis: design, implementation and evaluation. Future Internet 12(7). https://doi.org/10.3390/fi12070109
Martín-Herrera I, Belda JPM (2021) Efectos de la desinformación sobre la imagen de marca y reacción de tres empresas del sector alimentario ante una crisis de comunicación: Starbucks, Mercadona y Burger King. Obra Digit (20). https://doi.org/10.25029/od.2021.285.20
Mehmood K, Kautish P, Shah TR (2024) Embracing digital companions: unveiling customer engagement with anthropomorphic AI service robots in cross-cultural context. J Retail Consum Serv 79:103825. https://doi.org/10.1016/j.jretconser.2024.103825
Article Google Scholar
Men LR, Zhou A, Jin J, Thelen P (2023) Shaping corporate character via chatbot social conversation: impact on organization-public relational outcomes. Public Relat Rev 49(5):102385. https://doi.org/10.1016/j.pubrev.2023.102385
Article Google Scholar
Mikušová M, Horváthová P (2019) Prepared for a crisis? Basic elements of crisis management in an organisation. Econ Res Ekon Istraž 32(1):1844–1868. https://doi.org/10.1080/1331677X.2019.1640625
Article Google Scholar
Murphy J, Gretzel U, Pesonen J (2019) Marketing robot services in hospitality and tourism: the role of anthropomorphism. J Travel Tour Mark 36(7):784–795. https://doi.org/10.1080/10548408.2019.1571983
Article Google Scholar
Ozanne LK, Ballantine PW, Mitchell T (2020) Investigating the methods and effectiveness of crisis communication. J Nonprofit Public Sect Mark 32(4):379–405. https://doi.org/10.1080/10495142.2020.1798856
Article Google Scholar
Panda G, Upadhyay AK, Khandelwal K (2019) Artificial intelligence: a strategic disruption in public relations. J Creat Commun 14(3):196–213. https://doi.org/10.1177/0973258619866585
Article Google Scholar
Pantano E, Pizzi G (2020) Forecasting artificial intelligence on online customer assistance: evidence from chatbot patents analysis. J Retail Consum Serv 55:102096. https://doi.org/10.1016/j.jretconser.2020.102096
Article Google Scholar
Pasquale F (2015) The blackbox society: the secret algorithms that control money and information. Harvard University Press
Persons B, Jain P, Chagnon C, Djamasbi S (2021) Designing the empathetic research IoT Network (ERIN) chatbot for mental health resources. In: Nah FF-H, Siau K (eds) HCI in business, government and organizations. Springer International Publishing, pp 619–629
Prahl A, Goh WWP (2021) Rogue machines” and crisis communication: When AI fails, how do companies publicly respond?. Public Relat Rev 47(4):102077. https://doi.org/10.1016/j.pubrev.2021.102077
Article Google Scholar
Qadir J, Ali A, ur Rasool R, Zwitter A, Sathiaseelan A, Crowcroft J (2016) Crisis analytics: big data-driven crisis response. J Int Humanit Action 1(1):12. https://doi.org/10.1186/s41018-016-0013-9
Article Google Scholar
Rahman MS, Bag S, Hossain MA, Abdel Fattah FAM, Gani MO, Rana NP (2023) The new wave of AI-powered luxury brands online shopping experience: the role of digital multisensory cues and customers’ engagement. J Retail Consum Serv 72:103273. https://doi.org/10.1016/j.jretconser.2023.103273
Article Google Scholar
Schniter E, Shields TW, Sznycer D (2020) Trust in humans and robots: economically similar but emotionally different. J Econ Psychol 78. https://doi.org/10.1016/j.joep.2020.102253
Schuetzler RM, Grimes GM, Scott GJ (2020) The impact of chatbot conversational skill on engagement and perceived humanness. J Manag Inf Syst 37(3):875–900. https://doi.org/10.1080/07421222.2020.1790204
Article Google Scholar
Sheehan B, Jin HS, Gottlieb U (2020) Customer service chatbots: anthropomorphism and adoption. J Bus Res 115:14–24. https://doi.org/10.1016/j.jbusres.2020.04.030
Article Google Scholar
Taneja SL, Pryor MG, Sewell S, Recuero AM (2014) Strategic crisis management: a basis for renewal and crisis prevention. https://api.semanticscholar.org/CorpusID:167648415
Telang A, Deshpande A (2016) Keep calm and carry on: a crisis communication study of Cadbury and McDonalds. Manag Mark 11(1):371–379. https://doi.org/10.1515/mmcks-2016-0003
Article Google Scholar
The New York Times (2014) Chief of Abercrombie & Fitch is stripped of chairman title. The New York Times. https://www.nytimes.com/2014/01/29/business/chief-of-abercrombie-fitch-is-stripped-of-chairman-title.html
Trivedi J (2019) Examining the customer experience of using banking chatbots and its impact on brand love: the moderating role of perceived risk. J Internet Commer 18(1):91–111. https://doi.org/10.1080/15332861.2019.1567188
Article Google Scholar
Victor D, Stevens M (2017) United Airlines passenger is dragged from an overbooked flight. The New York Times. https://www.nytimes.com/2017/04/10/business/united-flight-passenger-dragged.html
Xiao Y, Yu S (2025) Can ChatGPT replace humans in crisis communication? The effects of AI-mediated crisis communication on stakeholder satisfaction and responsibility attribution. Int J Inf Manag 80:102835. https://doi.org/10.1016/j.ijinfomgt.2024.102835
Article Google Scholar
Yamanoue T (2017) Monitoring servers, with a little help from my bots. In: Proceedings of the 2017 ACM SIGUCCS Annual Conference 17:173–180. 10.1145/3123458.3123461
Yoon VY, Hostler RE, Guo Z, Guimaraes T (2013) Assessing the moderating effect of consumer product knowledge and online shopping experience on using recommendation agents for customer loyalty. Decis Support Syst 55(4):883–893. https://doi.org/10.1016/j.dss.2012.12.024
Article Google Scholar
Zepeda M (2012) The Rat’s life environment for research and education [Bachelor of Science Thesıs]. Budapest University of Technology and Economics
Zhang T, Feng C, Chen H, Xian J (2022) Calming the customers by AI: investigating the role of chatbot acting-cute strategies in soothing negative customer emotions. Electron Mark 32(4):2277–2292. https://doi.org/10.1007/s12525-022-00596-2
Article Google Scholar
Zhu W, Lu S, Huang P, Hu X, Wei J (2023) Predictors of coping behavior during the COVID-19 pandemic: evidence from China. J Contingencies Crisis Manag 31(4):797–808. https://doi.org/10.1111/1468-5973.12481
Article Google Scholar
Download references
The author(s) received no financial support for this study’s research, authorship, and/or publication.
Public Relations, Tokat Gaziosmanpaşa University, Tokat Province, Turkey
Murat Seyfi & Hıdır Polat
Public Relations, Marmara University, Istanbul, Turkey
Yeliz Kuşay & Deniz Güven
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
Murat Seyfi conceptualised the study and designed the research. Hıdır Polat conducted the data collection and analysis processes. Yeliz Kuşay and Deniz Güven prepared the figures/tables and carried out the visual layout. Murat Seyfi and Hıdır Polat wrote the main text. All authors reviewed the final version of the manuscript and approved it for publication.
Correspondence to Murat Seyfi or Hıdır Polat.
The authors declare no competing interests.
We collected data between May and June of 2025. Prior to the study, ethical approval was obtained from the Ethics Committee of Tokat Gaziosmanpaşa Üniversitesi (Date: 29.05.2025, Decision No: 01-121). The research process was conducted in accordance with the Declaration of Helsinki, and informed consent was obtained from all the participants.
The participant consent forms for the study were submitted to the Ethics Committee of Tokat Gaziosmanpaşa University and approved. Informed consent was obtained from all participants, providing them with information about the study and assuring them that their personal data would be safeguarded. The aim, methods, potential risks, and benefits of the research were clearly explained to each participant, and the participants voluntarily agreed to take part in the study. Participants were assured of their right to withdraw from the study at any time. All concerns were addressed prior to obtaining consent. During the data collection process, which took place in May and June 2025, each participant read the consent forms and signed them to confirm their acceptance. Participation was entirely voluntary, and interviews were conducted only with those who gave explicit consent. This consent process was witnessed by the authors. Participants’ confidentiality and anonymity were strictly protected throughout the research process.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Tables 1 and 2
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
Seyfi, M., Polat, H., Kuşay, Y. et al. AI-supported chatbots as triggers of potential communication crises: a phenomenological study of user experiences. Humanit Soc Sci Commun 13, 1222 (2026). https://doi.org/10.1057/s41599-026-07617-x
Download citation
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1057/s41599-026-07617-x
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Advertisement
Humanities and Social Sciences Communications (Humanit Soc Sci Commun)
ISSN 2662-9992 (online)
© 2026 Springer Nature Limited