- Determine the measure levels of public trust in the AL tool Grok in times of crisis.
- Separate civil faith towards AI vs traditional journalism.
- Hold back whether dependency on AI drives viewers off reporters.
- Address the affective and thought processes of an AI tool such as Grok.
- Extends understanding of trust in AI tools
- Deepends Media Dependency Theory connectivity to algorithmic communication
- Contributes to the journalism integrity debate
- Helps policymakers in making AI regulations
- Supports integration of AI in newsrooms
- Rarely: 46.8%
- Several times a week: 23.9%
- Once or twice: 17.4%
- Daily: 11.8%
- Moderate trust in Grok: 54%
- High trust: 18%
- Very high: 8%
- Low trust: 12%
- More reliable than journalists: 69.2%
- About the same: 13.9%
- Less reliable: 8.2%
- Not sure: 5.3%
- Agree/Strongly Agree: 60.3%
- Disagree/Strongly Disagree: 24.2%
- Neutral: 15.5%
- Journalists should work with AI, not be replaced:
- Agree/Strongly Agree: 64.7%
- Disagree/Strongly Disagree: 12.1%
- Neutral: 22.9%
- Mean = 2.90 (SD = 0.60)
- Range: approximately 1.50 to 4.50
- There was a noticeable number that said that they had witnessed wrong information produced by Grok.
- Many also said that Grok would show wrong information.
- AI as a Quick and Assisting Tool: Numerous respondents were grateful for the speed and straightforwardness of Grok, especially when rapid action is happening.
- Journalistic Interpretation is Still Required: Often, the respondents mentioned that AI did not have the context, emotional interpretation, field reporting, or moral judgment.
- Bias towards Hybrid Models: It is inherently biased towards the idea that AI should be used to assist journalists and not to substitute humans, which is viewed as the urge to balance automation and human decision-making.
- Trust in Grok is moderately high, and slightly higher than trust in journalists.
- Usage of Grok during the 2025 crisis was widespread, with nearly everyone using it at least once.
- Grok is perceived as more reliable than journalists by the majority of respondents.
- Dependency is moderate, indicating reliance but not overdependence.
- Trust in Grok does not displace trust in journalists; both function as separate trust pathways.
- Users express concerns about misinformation, suggesting a critical rather than blind trust.
- Most respondents favor a collaborative human, AI model, not full replacement.
Abstract
This study explores public trust in Grok, an AI fact-checking tool on X (Twitter), during the 2025 Pakistan-India war. Using Media Dependency Theory, it examines how Grok shaped trust, fact-checking behavior, and perceptions of journalistic authority. A mixed-methods survey of 380 respondents measured trust, dependency, and views on AI's role in crisis reporting. Findings show moderately high trust in Grok, moderate trust in journalists, and a perception that Grok was more reliable during the conflict. However, most participants favored human-AI collaboration over replacing reporters. Dependency on Grok was moderate and positively linked to trust. While respondents feared misinformation, they still valued AI for verification. Overall, the study suggests AI will not replace journalism but is becoming a parallel authority, influencing how audiences navigate uncertainty in conflict situations.
Keywords
Artificial Intelligence, Grok, Public Trust in AI, Journalism, Crisis Communication, Media Dependency Theory, Fact-Checking, Misinformation.
Introduction
Background of the Study
The environment of information in the whole world has been severely changed by the modern development of AI and its adaptation to social networks, as well as its application by citizens. The new online ecosystem provides not just consumers with the ability to take in information but also enables interactions with automation that generates, summarizes, and authenticates live news. As compared to the traditional media, which involved the services of professional gatekeepers, the current information environment is becoming increasingly reliant on the services of computational systems in the interpretation of reality to users. (Lewis et al., 2019)
The sphere of AI has evolved beyond mere fundamental automation, like the spell-check option or search ranking. Daily, AI-powered systems play an important role in the development of content, gathering data, and checking. The language created with AI, such as Grok, may seem journalistic and have a high level of fluency and responsiveness to context. It has therefore also been adopted by not only the journalists, but also as a direct source of information and fact-checking for the users.
This change is particularly essential in the case of a crisis. Wars, natural calamities, civil disturbances, and security catastrophes generate high information, internet traffic, and an order of anarchy. In these crisis situations, the internet viewers want to find solutions quickly, and in many cases, they are not keen on official verification but rather accessibility and speed. This urgency pushes the users towards the sources that seem to be authoritative and fast.
The crisis between Pakistan and India in 2025 helped to generate an information vacuum where people were overly dependent on digital networks and required access to fast information and fact-checking. News, claims, and counter-claims keep being disseminated in the digital spaces. During this type of crisis, the social media networks and their contextualized AI-powered tools gave people immediate explanations of the occurring events. As an in-platform informational AI tool, Grok also provided the users with explanations and fact checks without navigating them through the established channels of journalism.
The trust in traditional news media has declined in most countries, and the global surveys illustrate fairly low confidence in news outlets, as well as a negative dynamic in the trust in journalism compared to the last few years (Reuters Institute, 2024). The decline in people's trust in journalism is explained by the fact that people are inclined to think about the bias of politics in the media sphere, sensationality, misinformation, and ownership pressure in media organizations (Fawzi, 2021). It has been found that users tend to place more trust in the technical systems as they are seen as objective and apolitical, even when they are not transparent (Sundar, 2008). This paper examines the question of whether audiences placed more trust in Grok than in traditional media in the 2025 Pakistan-India crisis and how the algorithmic authority influences the fact-checked behavior in the circumstances of emotional pressure and information overload. With fake news flooding the media landscape, direct bias and conflicting reports that users can readily obtain through AI, Grok may find a spot in the market as a tool of quick fact-checking and supposedly unbiased reporting. The analysis of the effects of perception of speed and neutrality, and trust on relying on Grok in the Pak-India war 2025 is considered. It also examines did this dependency indicated a falling trust in conventional journalism, or could it be that accessibility and convenience were the sole choice. Finally, the study will seek to learn about the bargaining of trust between human institutions and AI tools in times of national and emotional ambiguity.
Problem Statement
Although the use of AI and its combination with digital networks is widely accepted, the lack of academic research in the field has been identified in the number of studies devoted to the topic of trusting AI tools in real-time crisis situations. Although many works have been conducted on the topic of algorithmic journalism, few studies have investigated how AI tools can transform rattles and flabbergast in terms of the confidence of the populace. The majority of available literature places more emphasis on newsroom orientation as opposed to audience behavior.
Artificial intelligence systems and devices are very likely to speak with high levels of linguistic confidence. This could give the feeling of certainty to information that is skewed or not comprehensive. In contrast to traditional journalism, where an ethical code is adhered to strictly, AI tools generate outputs that lack accountability frameworks, and in many cases anonymity of the source is also not known. This central difference is worrisome in terms of framing trust and whether viewers can wrongly confuse fluency and factual accuracy.
The main issue wrought by this was that Grok served a stabilizing purpose in concentration, or that it served to enhance dependence on computerized technologies. Assuming the credibility of AI without an empirical study of the public evaluation means a possibility of it being less objective and more rhetorical.
Purpose of the Study
This study seeks to:
Significance of the Study
This research:
Literature Review
Artificial Intelligence and Journalism
AI-based technology has grown at an outstanding pace in the last ten years, transforming the methods of production, curating, and distributing news. Artificial intelligence is now used in numerous newsroom services, such as creating automated reporting, detecting trends in big data, customizing news streams, suggesting news to visit/read depending on user behaviour, and evaluating the audience's interest/attention. Computational systems have become very useful in many organizations as they help with various routine reporting, be it financial reports, sports reports, or event reports. This is a general shift towards algorithmically-supported journalism, with AI working with the editors, shaping the newsroom and editorial outputs.
Diakopoulos (2019) believes that AI has come a long way beyond its first concept of being an optimization tool. It is currently actively involved in editorial management, having a role in determining what and who the nominations to a story will be, the tiering of headlines, and the information that ultimately is received by the audience. Algorithms determine the popularity of stories and the significance that various stories get through personalization and automated distribution. It marks a radical change in the power of journalism, in which editorial influence is being divided more and more among human participants and computers.
The old-fashioned mode of gatekeeping that was based on the newsroom ethics and human judgment has changed into a hybrid mode. Most of the filtering and prioritizing processes that were previously in the hands of journalists are now handled by algorithms. Carlson (2015) also writes that AI does not just assist with journalistic activities, but it is also efficient in storytelling practice. Since automated systems do not include moral reasoning, the issue of accountability gets more complicated when misinformation or bias are present in AI-generated news.
Whereas automation is fast and scalable, it lacks the investigative layers, context awareness, as well as ethical reflection that human journalism is more about. With the increased use of AI in news production, researchers have cautioned that excessive use of the automated systems could compromise transparency and editorial accountability. AI in journalism is not only that; it is an enormous change in the perception of authority, credibility, and responsibility in news coverage in the digital era.
AI in the role of a Verification Authority
In conventional journalism, fact-checking has always been correlated with institutional practices, editorial control, and professional values. Verification is not considered a single process, but a continuous process that requires cross-checking sources, document analysis, and review of evidence before being published. These practices assist news organizations in fostering legitimacy and social trust. Although spectators realise that journalists are capable of mistakes, their needs are openness, accountability, and readiness to rectify the error when it takes place.
According to Graves and Anderson (2020), professional fact-checking is inherent in a social organization that relies on mutual trust, instead of technical expertise. Journalists are educated to judge the credibility of the sources, assess the information in the framework of the social context, and think about the effect of misinformation publication. This interpretive position makes media more credible and makes journalism an institution of continuity, critique, and responsibility, rather than rapidity.
Verification systems that are powered by AI work in a completely different manner. Their power has no source in training, institutional norms, or human accountability. Rather, it is a result of their capacity to generate quick outputs and perfected responses. These systems may be convenient, but in most cases, it is not made known how they arrived at this response and what sources they use. The information selection, processing, and framing are frequently not well-known to users. This secrecy is a direct provocation to the standards of journalistic checking.
According to Tandoc and Maitra (2023), the fluent and confident language implemented by AI may contribute to giving a false impression of reliability. Since the reactions have tones similar to those of professional journalism, the users might think that the information could be true even when the factual foundation is not obvious. Such a dynamic promotes the performative credibility that the authors define as a type of trust that is determined by style and action instead of evidence.
As the use of AI tools increases, it is possible that trust will be narrowed in favor of news bodies and taken over by automation. The verification process turns into an individual goal instead of a group work of professionals. In this environment, AI tools start taking a symbolic position that was previously played by journalists, causing significant questions about the way the public is influenced by the establishment of a technologically delegated judgment.
Declining Trust in Journalism
Trust in journalism has been constantly declining throughout most sections of the globe due to the combination of political polarization, the change in technology, and the alteration of media space. The more societies are polarized, the more audiences are reporting news based on their ideological positions. Even when the reports contradict the individual beliefs, they are widely ignored as biased, even though the reports are actually true. This trend renders journalism to act as a consensus provider of truth because individuals take information with some skepticism, as opposed to being neutral.
This issue is augmented by the proliferation of digital fake news. There are unhealthy claims, rumors, and fake narratives that, through online platforms, will spread faster and widely. Such faulty accounts tend to spread far earlier than professional journalists can research events and provide perspective. Due to the mix of misleading content and credible news, the readers can hardly differentiate between reliable and unreliable news because they are located in the same places. This bewilderment puts the legitimacy of journalism as a good guide of what people know at risk.
Details provided in research conducted by the Reuters Institute (2024) reveal that although the number of people accessing news is greater than ever, the exposure to the news did not lead to a higher degree of trust. Quite to the contrary, the level of confidence is dropping, indicating that access to information does not necessarily lead to faith in its ability to be trusted. The report blames the trend on the saturation of competing voices on the internet, such as influencers, partisan pundits, and other media sources. All these streams of information make the environment very fragmented, where journalism struggles to retain its legitimacy.
As Strombeck and others (2020) put it, accuracy is not the only factor that makes people trust journalism, but it also depends on the perceptions of the institutions that govern the news being reported. Fairness, transparency, and independence are some of the factors that contribute to the creation of credibility. Distrust goes up when media houses are perceived to be politically biased or are driven by commercial interests. Skepticism of digital platforms, especially with regard to the problem of algorithms and echo chambers, tends to transfer into value systems with regard to journalism itself. This overlap of platform distrust and media distrust further adds to the relationship between audiences and the press, which leads to a wider environment of insecurity.
Trust in Automation and AI
The degree of trust in AI systems has been agreed upon
to be relative. Lee and See (2004) concede that AI trust develops due to the assessment of circumstantial demands by an audience. People are typically ready to trust machines and AI solutions when they need to use precision, efficiency, and reliability in performing tasks, such as a situations where the speed at which processes information takes priority. Simultaneously, users do not want to believe in cases when the work demanded a moral choice, moral judgment, or interpretation. This disparity demonstrates that automation is valued as an appropriate technology because of its technical advantages, not as a person-replacing tool in difficult or emotionally charged fields of operation.
Expanding upon this theory and idea, Longoni et al. (2019) demonstrate the tendency of discomfort in automated systems to increase, covering the situations when the choice concerns emotionally or socially important cases. According to their study, people consider such systems speedy and efficient, but they are unwilling to delegate authority when the decisions concern matters of health, personal relations, or morally involved cases. This is hesitation based on the view that AI tools cannot be empathetic, have moral judgment, and understand nuanced human situations. The increased the expressive stakes of a decision, the less confidence in automated systems, the more likely it is to be maintained, despite the technical competence of those systems. It is a pattern that can be used to justify the reason why it is possible that many people are hesitant about only using automated verification tools in times of confusion or crisis.
Crisis communication also has a distinctive position in the sense that factual integrity should be combined with emotional comfort. When facing events such as geopolitical struggles, natural calamities, or national crises, people are seeking something other than simple information. They need clarification, counsel, and psychological stability. AI applications, such as Grok, can provide both fast and organized evaluations of going statements, allowing for to reduction of information noise during periods when rumors and speculations are going viral. Simultaneously, such tools also tend to play a symbolic role: their assertive sound and the regularity of their responses may be perceived by the users as some form of stability or power. This dualism implies that AI systems can be used not only as technical tools to process information but also as emotional basal structures that can be perceived. Together with the effects, one can highlight the necessity to investigate the issue of AI authority in crises, not only concerning its precision and rapidity of the reaction, but also in regard to how these mechanisms influence the views, confidence, and sense of security among people during the time of increased uncertainty.
Automation Bias
According to Mosier and Skitka (1999), automation bias refers to a behavior where people tend to believe machine answers in case there are evident indications that the system is in error. It occurs mostly at those instances where there is time pressure, lots of work, or complex information. The less people are tired of the mental load, the more they desire automated suggestions to be agreed with without hearing them. At such times, automation is not only useful. It also begins to play an overriding role in decision-making that the user is sometimes likely to be unaware of the extent of its sway.
Skitka et al. (2000) build on this research by demonstrating that individuals generally do not doubt or disbelieve machine recommendations despite having the unambiguous risks of taking such a course of action. According to the research that they have conducted, as soon as automation demonstrates a constant pattern of doing a good job, they rely on it and start neglecting to critically analyze the outcomes. This reliance persists despite the inconsistency of the automated recommendation with another source of information, and it has very severe repercussions for erring. According to the researchers, individuals can afford to trust automated systems as it is easier and more comfortable to use automated systems than to undertake subsequent analysis with potential avoidable falsehood.
These tendencies are even more pronounced in moments of crisis. Emergencies establish circumstances under which individuals have to make decisions fast, but they have to contend with panic, misunderstanding, and anxiety. In such a situation, people can view automated systems as being more likely or sturdier than human resources, particularly when data is questionable or contradictory. Such tools as Grok are thus free to assume more competence in times of crisis, not only because they give a rapid response but simply because they appear driven and reliable. This begs some important questions regarding the basis of trust in AI validation during crises, as grounded in a personal belief in the system itself, or is driven by a stress-induced reaction and a lack of mental capacity. The concept of automation bias will thus be important in the evaluation of how AI-enabled verification systems influence the process of decision-making by the people in times of instability.
Fake News and Virality
Studies of online media have always shown that misinformation does not respond to verified news in the same manner, particularly concerning the speed of spreading. Vosoughi et al. (2018) established that fake news reaches a broader audience faster than verified information on social networks. Their analysis unveiled that emotionally appealing content like fearful, angering, or surprising stories spread more quickly as they are seen to arouse psychologically strong responses. These emotional reactions predispose the users towards sharing information without stopping to consider whether it is accurate or not. Consequently, fake news spreads faster in the online ecosystem, not due to any truth value attainment, but due to its entertaining and sensational nature. This dynamic can be particularly problematic in situations of crisis when emotions are increased, and users are actively in search of clarity.
Allcott and Gentzkow (2017) also prove that misinformation is not just highly disseminated but can also impact ideology and behavior, specifically in politically different situations. Their results demonstrate that misleading political allegations affect attitudes and voting intentions, which implies that incorrect information may be practically effective in affecting the decision-making process of the population. Misinformation is used to fill interpretive gaps in high polarization settings and provide individuals with emotionally gratifying explanations even when the latter are not founded on facts.
When a crisis comes, the stakes of this dynamic increase tremendously. When in a state of uncertainty, these people are more likely to be concerned with what is immediate than with evaluations. The pressure to explain the events occurring to them makes them vulnerable to anything that can provide them with an easy solution. In case digital platforms show the arbitration and assertiveness of the algorithm systems to users, this immediacy can be perceived as a gesture of authority. Verification tools powered by AI, as well as Grok, will consequently become beneficial not only because of their ability to respond quickly but also because of the psychological comfort the use of those tools can provide. During such times, algorithmic power becomes more potent since it is guaranteed of immediate clarity when human professionals or reporters are still going out there to collect data. The analysis of this change is critical to analyzing the reasons why viewers can afford automated validation at times of increased anxiety and perplexity.
Media Dependency Theory
The Media Dependency Theory can be used to explain the results of individuals resorting to media in cases of uncertainty. According to Ball, Rokeach, and DeFleur (1976), individuals turn to the media to satisfy three fundamental needs, namely understanding the world, their role in social, political environments, and entertainment. This reliance is heightened in times of unsteadiness, e.g., in times of political warfare or natural calamities, or social disturbance. Breakdowns in familiar routines call on increased dependence on information systems that may assist people in trying to make sense out of the emerging events. Media thus forms an important source of decreasing the uncertainty and regaining the sense of control.
Lowrey (2004) takes the Media Dependency Theory into the digital age by demonstrating how the emergence of new technologies transforms the forms of individual reliance on information systems. With the rise of digital platforms dominating the daily news content one by one, the habit of relying solely on the traditional news sources was discarded slowly by surely. They began instead to resort to interactive and tailor-made systems, which sift and sort information for them. These media do not just repeat the news. They also determine the order in which stories are shown at the top, whether an information source is reliable or not, and the rate at which people receive updates. The dependency in such an environment is even more complicated since the informational power is commonplace among journalists, algorithms, and the digital interfaces that instruct the audience to operate in a particular way.
AI technologies are described by Lowrey as dependency nodes. They control the informational influence by determining what news is the most significant, present rapid clarification, and answer the questions posted by users. In a crisis or when an event becomes uncontrollable, individuals might seek the help of AI equipment due to the ability to gain a clearer picture, in a time when journalists will continue with the procedure of fact-checking. These tools might be especially trusted, as the answers provided by AI are fast and sound, and confident. This increases dependence despite the fact that the users may not know the complete mechanisms that the systems use to choose or curate information.
These trends are indicators that AI is no longer an accessory to the information world. It is becoming influential in the way individuals perceive events, make judgments, and the manner in which they engage in uncertainty. Under a high level of pressure, the fact that AI can process information quickly and that the sense of objectivity emerges makes it have a type of authority that users might easily follow without serious questions. Subsequently, AI tools are increasingly becoming involved in informing how the masses process the meaning of crises and other emotionally charged situations.
AI and Authority Perception
Human-computer contact studies have demonstrated that human beings tend to react towards computerized systems in the same manner as they react towards other human beings. Nass and Moon (2000) prove that even when an individual is well aware of the fact that a mechanism is a machine, he or she implicitly transfers the rules of social behavior to machines. As a system talks naturally, adheres to the rules of conversation, or puts across information in a well-assured manner, users are likely to think of social behaviors like competence, politeness, or trustworthiness. This implies that power is not generated merely through accuracy, but how it is interrelated as well as the elicits it provides regarding expertise.
In the perceived legitimacy, fluent communication is especially significant. When AI is able to give clear and well-structured explanations, it may give the user the sense that a form of knowledge is inherent. They, in most instances, fail to question how the system got to its conclusion or what limitations it may have. This type of trust, based on the impression made, can bring AI systems to a level that they can be compared to the human level of expertise, despite the fact that the process of the results of AI outputs and those of humans is completely different.
This is further reinforced by perceived neutrality. Most users believe that AI systems are neutral since they are not human and hence are not driven by political or self-interest. This supposition could turn the AI suggestions into more unbiased or truthful than the data provided by journalists, institutions, or colleagues. Consequently, AI authority is not just a norm within the performance but within psychological comfort as well. Individuals can trust AI since it seems to be unbiased, predictable, and unemotive, which is comforting in times of uncertainty.
When combined, AI systems will be able to gain a social and epistemic authority outside of their technical capacities. This type of perception of AI by the users is important in the consideration of the growing role of AI tools in high-stakes informational environments, especially in the context of a crisis, wherein audiences are looking not only to find clarity but also a sense of stability.
Crisis Communication
Palen and Liu (2007) note that crisis situations shift audience priorities toward quick responses rather than accuracy. When information is rare and uncertainty is high, people share updates fast in an effort to lessen confusion or help others make sense of unfolding events. As a result, distribution often replaces careful evaluation or deliberation. In these conditions, fast-circulating messages gain influence simply because they arrive first, shaping public perception before verified information becomes accessible.
Research Gap
Research has explored algorithmic news curation,
but not audience trust in AI as a verification authority during international conflict. No large-scale studies assess Grok's role in news verification and public trust in it. This study addresses that void.
Theoretical Framework
Media Dependency Theory provides the conceptual foundation for understanding why audiences increasingly relied on Grok during the Pakistan-India crisis of 2025. According to Ball-Rokeach and DeFleur, dependence on informational systems intensifies when social uncertainty or perceived threat increases. During crises, audiences seek three core functions from media: surveillance to monitor unfolding events, orientation to interpret what those events mean, and reassurance to manage emotional stress. Grok appears to address these needs all at the same time by providing fast updates, background information, and a confident tone that will help minimize the ambiguity. The more they use Grok to find clarification and seek emotional groundedness, the more addicted these users become to the system. This increased dependency makes Grok seem more authoritative, and audiences view its products not as information, but as advice. Media Dependency Theory thus explains how an AI tool can develop to impact the citizens and trust formation at a time when a nation is in doubt.
Research Questions
RQ1: How often did users use Grok?
RQ2: Is there an accuracy and neutrality difference in Grok perception?
RQ3: Was there diminishing trust towards journalism?
RQ4: What were the strongest risks related to the users?
RQ5: Did Grok displace journalists psychologically?
Research Design
This research paper will assume a mixed-methods research design, which will combine the elements of quantitative and qualitative research to investigate the levels of trust toward Grok by the people during the 2025 Pakistan-India crisis. The primary part of the research is a structured survey that will describe the following: how individuals will utilize Grok, how believable it will be, and the degree how which individuals will optionally consult it to determine the truth during the Pak-India war 2025. In line with this, the survey has additional open-ended items, whereby the participants are encouraged to provide whatever they encountered and what they considered useful, their concerns, as well as how they came up with certain judgments concerning trust. By integrating these two methods, the study will be able to establish bigger trends among its users as well as understand how individuals form their own understandings of AI-based technologies that act within social media systems at a more individual and intimate level. This combination approach is useful in both exposing the frequency of the times that people turn to Grok and their rationales of doing so, as well as whether they depend on it or not, or compare it to more traditional journalistic outlets.
Sampling
The purposive sampling method was used to allow the researcher to identify all individuals who had direct experience with the use of Grok within the time frame of the study. Since the study is geared towards exploring the role as perceived by real users of Grok as an information authority, the participants were chosen in relation to their being conversant with the tool. Such a method will make sure that respondents will be able to seriously assess the performance of Grok, his credibility, and influence at the event. Although the non-probability sampling reduces generalizability, it is suitable in exploratory studies that aim at a particular group of users.
Instrument
The research gathered data by way of a 30-item online questionnaire, which comprised closed and open-ended questions. The tool had Likert-scale items which were evaluating trust, perceived neutrality, accuracy, emotional reassurance, and dependency. Other items evaluated the frequency of use, dependency on Grok as opposed to the use of journalists, and the clarity offered by the person during the crisis. Open-ended remarks authorized the respondents to discuss issues, interpretive experiences, and anticipations of AI validation tools.
Analysis
The descriptive statistics of means, frequency distributions and there will also be a composite trust index to evaluate general confidence in Grok as a whole will be applied to quantitative data analysis. Correlational analysis shall be used to examine the relationships among trust and usage patterns, and dependency indicators. Thematic analysis will be applied to qualitative responses, which will be coded repeatedly to determine the designs of perceived authority, emotional assurance, anxieties about the accuracy, and perceptions of AI power during crisis moments. The combination of such results will enable the research to introduce a comprehensive description of the dynamics of perception and trust in the media.
Ethical Considerations
The research guarantees ethical quality by ensuring data was collected in an anonymous manner, and those who gave informed consent are the ones who participated in the research. No names or identifiable data are collected, and a respondent is free to drop out. The study will be voluntary, and will follow ethical principles of confidentiality, respect, and transparency in research involving human beings.
Findings
Sample Characteristics
One hundred and eighty responses were collected and analyzed. The sample was highly youthful and digital in nature. Almost fifty per cent of the respondents were aged between 18-25 (48.2%), 26-35 (44.7%), and a minor percentage aged 36-45 (7.1%). Females were 55.8, males 43.7, and a minor percentage did not want to reveal their gender.
The interest of the respondents in political and conflict-related content was high: 36.1% said that they pay much attention to such news very frequently, and 36.3% frequently. The places of news consumption were dominated by digital platforms, which aligns with the tendencies of online verification behavior in modern media landscapes.
Awareness and Use of Grok
Grok awareness was unusually high: more than 99
Of them noted that they had heard of this tool, and 97.1 of them had used it at least once. In the Pakistan-India crisis 2025, 96 percent reported relying on Grok specifically to confirm the news, making it evident that the vast majority of people are trying to use AI-generated news in situations when they are highly stressed.
Use frequency varied:
These results suggest that while almost all participants used Grok, the intensity of usage ranged from light to habitual.
Trust in Grok and Trust in Journalists
Seven items, which comprised perceptions of accuracy, neutrality, clarity, and reliability, were used to create a composite Grok Trust Index. The index generated a median of 3.37 (SD = 0.50) based on a scale of 1-5, which is the median degree of trust in Grok.
Specifically, concerning general trust:
Trust in Journalists, in its turn, produced a mean of 2.99 (SD = 0.82). A large proportion of the respondents expressed moderately high trust (60.5%), and the proportion of respondents who expressed high trust (very high) is very slim (4.7%).
Correlation analysis revealed no significant correlation between the trust in Grok and the trust in journalists (r ≈ 0.01). This implies that the trust towards AI is not merely replacing and/or reducing the trust in traditional journalism, rather users appear to be judging these information sources separately.
Perception of Grok Relative to Journalists
The respondents were to directly compare Grok to journalists:
Therefore, close to seven out of ten participants rated Grok more reliable than human journalists in times of crisis. The perceptions of AI as a substitute or complement of journalism were both positive and negative, but tended to the latter:
AI should replace journalists for fact-checking:
These results show that although a relatively high number of respondents believe that AI-driven fact-checking is desirable, more of them believe in the hybrid forms of human-AI, which implies that they appreciate the presence of human involvement in the field of journalism.
Dependency on Grok
The Grok Dependency Index added the usage frequency to the reliance verification agreement. The index produced:
This is an indicator of moderate dependence among overall respondents who used Grok regularly but were not wholly dependent on it to the level of substituting other sources.
A small to moderate positive correlation between the Grok Trust Index and Dependency Index showed that the two were correlated (r ≈ 0.22). This implies that, despite the fact that the higher the trust, the higher the dependency, the connection between them is not strong enough to indicate blind trust; users can depend on Grok because of convenience even when trust is moderate.
Concerns About Accuracy and Misinformation
Although in general, the evaluation is positive, the participants noted the awareness of the AI limitations:
These issues co-exist with high usage numbers, which means that users can use Grok as a first-check device in place of a truth-determining tool.
Nevertheless, the respondents were not pessimistic regarding its future use: most of them stated that they would use Grok again in another crisis.
Open-Ended Responses
The qualitative question to be analysed, i.e., Can AI replace journalists? Why /why not? was found to bring out three themes:
Summary of Findings
The findings indicate that, combined, there is an indication that:
Discussion
The present research findings indicate that Grok has become a major point of verification when the Pakistan-India crisis occurred in 2025, and it is through such crises that audiences managed to work through uncertainty, as well as have different information online and make sense of it. As predicted by the Media Dependency Theory (Ball-Rokeach and DeFleur, 1976), people who had high information scarcity and psychological strain were more predisposed to fast and convenient sources. The instant reaction of Grok and its perceived non-taking stance made it a formidable negation that could have been adapted as an alternative to slower verification systems that were based on a department.
Among the most notable results is that an almost equal number of ten respondents felt that Grok was more trustworthy than journalists. This is a departure from the traditional forms of authority, where journalists were the only ones to do gatekeeping and verification. Rather, it appears that the concept of algorithmic authority has been brought to the norm, especially with the younger, digitally savvy users. This can be manifested in the moderate-high Grok Trust Index (M = 3.37), whereas marginally low trust in journalists (M = 2.99) can indicate that mistrust in mainstream media has not yet diminished. The absence of the connection between the two trust measures (r = 0.01) suggests, however, that trust in AI is not substituting the trust in journalists but acts as an alternative credibility route.
Another point of dual attitude to AI and journalism is mentioned in the survey as well. Although 60.3% of people said that AI might replace journalists in the aspect of verification, a more significant percentage (64.7) said that journalists should work together with AI and not be replaced. This trend indicates that the users perceive AI as productive and unbiased, yet they are still aware that human judgment and situational knowledge, and moral reasoning cannot be substituted in journalism. These results are consistent with research indicating that people are willing to leave technical jobs to machines, but emotionally, much or interpretive jobs are left to machines (Longoni et al., 2019).
Patterns of dependency also support this reading. Despite a very high level of awareness and use of Grok, the level of the index of Grok Dependency is (M = 2.90), which shows that it is not highly dependent but rather moderately dependent. The low value of the trust and dependency correlation (r ≈ 0.22) indicates that users are not putting their faith in Grok unquestioningly, but dependency is influenced by convenience and the necessity to get instant clarity in case of crisis.
Issues with misinformation are congruent. Participants have marked the cases when Grok gives wrong or incomplete information and shows concern about the possibility of AI, unknowingly, propagating misinformation. These concerns did not, however, have a significant impact on trust or intent to use in the future. This mix of apprehension and dependence points to the developing form of the hybrid trust, wherein users are conscious of the limitations of AI, but there is a desire to use it when time-constrained and more accurate than conventional media is too sluggish or partial.
Finally, the results indicate that journalism was not replaced, but redesigned the information ecosystem as proposed by Grok. The competency AI adopted was of a primary verifier with a journalism secondary contextualizer, which can redefine future conflict crisis communication.
Recommendations and conclusions.
This paper has discussed the ways in which the population would use Grok, an artificial intelligence verification application, in the 2025 war between Pakistan and India, and how these perceptions of trust and journalistic power would be influenced by that reliance. The findings indicate that Grok had a significant role in assisting viewers to maneuver in an ever-changing crisis context. Grok was seen by users as speedy, impartial, and convenient, and reinforced confidence in AI-powered fact-checking and triggered moderate dependency. Nevertheless, the mistrust of AI did not reflect in mistrust of journalists, which shows that the information needs of the two systems are different.
The implications of these findings are the following implications:
AI is emerging as a major fact-checking platform.
The AI tools can be more relevant during war scenarios than journalists building the first frame of truth.
Journalism is not a thing of the past but has to evolve.
The users are willing to use a hybrid mode whereby the journalists collaborate with the AI in enhancing the speed and accuracy of verification.
AI literacy has to be incorporated in media literacy.
The complexity of AI systems will require curriculum design to deal with the way AI tools, such as Grok, operate, where they go wrong, and how to be skeptical about the information presented by a machine.
The crisis communication systems should be redesigned.
Newsrooms, government regimes, and digital platforms are advised to incorporate AI in their verification process, as well as to come up with transparency, accountability, and protection against algorithmic fallacies.
Altogether, the use of AI fact-checking tools is not completely displacing journalists but modifying the trust ladder and changing the way of determining the truth in digital calamities. The future does not consist of deciding whether AI or journalism will prevail, but in building systems to have both to reinforce each other.
Further Research and Limitations.
Despite the fact that this research offers highly appreciated information concerning the role of AI checking tools in times of a crisis, it has a number of limitations that are to be considered. First, the sample was driven by a particular demography of online platforms, and it was based on the concept of purposive sampling that can overrepresent digitally active and younger individuals. This restricts the application to less digital and less accessible populations. Probability-based sampling should be used in the future to help obtain a more representative demographic profile.
Second, the research concerned only one X AI tool of Grok in a particular geopolitical crisis. Although the case provides an abundance of information to study, it is suggested that further research examines various AI systems, such as ChatGPT, Gemini, and Bard, to assess whether the trust pattern varies depending on whether this system is platform-based, interface-based, or even transparent. External validity would also be enhanced by making cross-national comparisons across various crises.
Third, self-reported data were used in the study, thereby subjecting the study to recall bias and social desirability. There are chances that the respondents will exaggerate their critical view about AI or minimise the use of traditional media. The next generation of work can be based on behavioral logs, eye-tracking, or digital traces that are going to be used to observe the behavior of verification in real-time.
Additionally, more insightful qualitative designs, including interviews, focus groups, or ethnographic research, can be used to complement insights into how users experience emotions and thoughts in the AI-generated information in the context of crisis situations.
In spite of these drawbacks, the research forms a very significant groundwork to grasp the notion of algorithmic trust, AI-based information dependency, and the new trends in artificial intelligence and journalism.
Tables
Table 1 Demographic Characteristics of Respondents (N = 380)
Table 2
| Variable | Category | Frequency (n) | Percentage (%) |
| Age Group | 18-25 | 183 | 48.2 |
| 26-35 | 170 | 44.7 | |
| 36-45 | 27 | 7.1 | |
| Gender | Male | 166 | 43.7 |
| Female | 212 | 55.8 | |
| Prefer not to say | 2 | 0.5 | |
| News Following Frequency | Very often | 137 | 36.1 |
| Often | 138 | 36.3 | |
| Sometimes | 77 | 20.3 | |
| Rarely/Never | 28 | 7.4 |
Descriptive Statistics for Trust and Dependency Index
Table 3
| Scale / Variables | Mean (M) | SD | Min | Max |
| Grok Trust Index | 3.37 | 0.50 | 1.83 | 5.00 |
| Trust in Journalists | 2.99 | 0.82 | 1.00 | 5.00 |
| Grok Dependency Index | 2.90 | 0.60 | ~1.50 | ~4.50 |
Comparison of Reliability: Grok vs Journalists
Table 4
| Response | Frequency (n) | Percentage (%) |
| More reliable than journalists | 263 | 69.2 |
| About the same | 53 | 13.9 |
| Less reliable than journalists | 31 | 8.2 |
| Not sure | 20 | 5.3 |
Attitudes Toward AI Replacing or Supporting Journalists
Attitudes Toward AI Replacing or Supporting Journalists
| Statement | Agree (%) | Neutral (%) | Disagree (%) |
| AI should replace journalists for fact-checking | 60.3 | 15.5 | 24.2 |
| Journalists should work with AI, not be replaced | 64.7 | 22.9 | 12.1 |
References
- Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of Economic Perspectives, 31(2), 211–236. https://doi.org/10.1257/jep.31.2.211
- Ball-Rokeach, S. J., & DeFleur, M. L. (1976). A dependency model of mass-media effects. Communication Research, 3(1), 3–21. https://doi.org/10.1177/009365027600300101
- Carlson, M. (2015). The robotic reporter: Automated journalism and the redefinition of labor, compositional forms, and journalistic authority. Digital Journalism, 3(3), 416–431. https://doi.org/10.1080/21670811.2014.976412
- Diakopoulos, N. (2019). Automating the news: How algorithms are rewriting the media. Harvard University Press.
- Fawzi, N. (2021). Concepts, causes, and consequences of trust in news media: A literature review and framework for analysis. Annals of the International Communication Association, 45(2), 154–169. https://doi.org/10.1080/23808985.2020.1847892
- Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1), 50–80. https://doi.org/10.1518/hfes.46.1.50.30392
- Lewis, S. C., Guzman, A. L., & Schmidt, T. R. (2019). Automation, journalism, and human–machine communication: Rethinking roles and relationships. Digital Journalism, 7(4), 409–427. https://doi.org/10.1080/21670811.2019.1577147
- Longoni, C., Bonezzi, A., & Morewedge, C. K. (2019). Resistance to medical artificial intelligence. Journal of Consumer Research, 46(4), 629–650. https://doi.org/10.1093/jcr/ucz013
- Lowrey, W. (2004). Media dependency during a large-scale social disruption: The case of September 11. Mass Communication and Society, 7(3), 339–357. https://doi.org/10.1207/s15327825mcs0703_5
- Mosier, K. L., & Skitka, L. J. (1999). Human decision makers and automated decision aids: Made for each other? In M. S. Helander, T. K. Landauer, & P. V. Prabhu (Eds.), Handbook of human-computer interaction (2nd ed., pp. 2017–2038). Elsevier. https://doi.org/10.1016/B978-044481862-1.50103-6
- Nass, C., & Moon, Y. (2000). Machines and mindlessness: Social responses to computers. Journal of Social Issues, 56(1), 81–103. https://doi.org/10.1111/0022-4537.00153
- Palen, L., & Liu, S. B. (2007). Citizen communications in crisis: Anticipating a future of ICT-supported public participation. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 727–736. https://doi.org/10.1145/1240624.1240736
- Reuters Institute for the Study of Journalism. (2025). Digital news report 2025. University of Oxford. https://reutersinstitute.politics.ox.ac.uk/
- Skitka, L. J., Mosier, K. L., & Burdick, M. (2000). Accountability and automation bias. International Journal of Human-Computer Studies, 52(4), 701–717. https://doi.org/10.1006/ijhc.1999.0349
- Strömbäck, J., Tsfati, Y., Boomgaarden, H., Damstra, A., Lindgren, E., Vliegenthart, R., & Esser, F. (2020). News media trust and its impact on media use: Toward a framework for future research. Annals of the International Communication Association, 44(2), 139–156. https://doi.org/10.1080/23808985.2020.1755338
- Sundar, S. S. (2008). The MAIN model: A heuristic approach to understanding technology effects on credibility. In M. J. Metzger & A. J. Flanagin (Eds.), Digital media, youth, and credibility (pp. 73–100). MIT Press.
- Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359(6380), 1146–1151. https://doi.org/10.1126/science.aap9559
Cite this article
-
APA : Paracha, T., Amjad, W., & Shah, Q. A. (2025). Grok and Public Trust: A Study of Algorithmic Authority in the 2025 Pakistan-India Crisis. Global Social Sciences Review, X(IV), 44-57. https://doi.org/10.31703/gssr.2025(X-IV).04
-
CHICAGO : Paracha, Tayyaba, Waqas Amjad, and Qasim Ali Shah. 2025. "Grok and Public Trust: A Study of Algorithmic Authority in the 2025 Pakistan-India Crisis." Global Social Sciences Review, X (IV): 44-57 doi: 10.31703/gssr.2025(X-IV).04
-
HARVARD : PARACHA, T., AMJAD, W. & SHAH, Q. A. 2025. Grok and Public Trust: A Study of Algorithmic Authority in the 2025 Pakistan-India Crisis. Global Social Sciences Review, X, 44-57.
-
MHRA : Paracha, Tayyaba, Waqas Amjad, and Qasim Ali Shah. 2025. "Grok and Public Trust: A Study of Algorithmic Authority in the 2025 Pakistan-India Crisis." Global Social Sciences Review, X: 44-57
-
MLA : Paracha, Tayyaba, Waqas Amjad, and Qasim Ali Shah. "Grok and Public Trust: A Study of Algorithmic Authority in the 2025 Pakistan-India Crisis." Global Social Sciences Review, X.IV (2025): 44-57 Print.
-
OXFORD : Paracha, Tayyaba, Amjad, Waqas, and Shah, Qasim Ali (2025), "Grok and Public Trust: A Study of Algorithmic Authority in the 2025 Pakistan-India Crisis", Global Social Sciences Review, X (IV), 44-57
-
TURABIAN : Paracha, Tayyaba, Waqas Amjad, and Qasim Ali Shah. "Grok and Public Trust: A Study of Algorithmic Authority in the 2025 Pakistan-India Crisis." Global Social Sciences Review X, no. IV (2025): 44-57. https://doi.org/10.31703/gssr.2025(X-IV).04
