Between Trust and Risk: Understanding the Conditional Acceptance of Artificial Intelligence

Abstract

Artificial Intelligence (AI) is rapidly transitioning from a specialized technology to an everyday socio-technical infrastructure, yet public acceptance remains shaped by a trade-off between perceived benefits and risks. This study examines how individuals from varied demographic and professional backgrounds perceive, use, and evaluate AI-enabled systems using a mixed-method research design. A bilingual (English/Arabic) online survey ($N=115$) captured demographics, awareness, usage patterns, perceived impact, self-assessed understanding, domain-specific trust, concerns, and attitudes toward regulation, complemented by open-ended reflections. In parallel, semi-structured face-to-face interviews provided deeper insight into AI conceptualization, lived experiences, trust boundaries, and conditions for acceptable use. Quantitative results show frequent AI engagement embedded in daily life, with strong domain dependence in trust: education is the most trusted domain, whereas healthcare and finance attract substantially lower trust. Prominent concerns include overreliance (“brain rot”), privacy and data misuse, job displacement, and misinformation. Support for stronger AI regulation is high, indicating that governance is viewed as a prerequisite for sustainable adoption rather than a constraint on innovation. Qualitative findings triangulate these results, revealing a pattern of conditional acceptanceunderstood as the simultaneous valuation of AI’s practical utility alongside the imposition of explicit trust prerequisites whereby participants value AI for productivity and learning support while emphasizing confidentiality, transparency, human oversight in high-stakes contexts, and clear boundaries to mitigate misuse and erosion of human judgment. The study offers empirically grounded insights for policymakers, educators, and industry stakeholders into how AI acceptance is negotiated through utility, literacy, perceived risk, and expectations of accountability.

1. Introduction

Artificial Intelligence (AI), once limited to science fiction and cartoons, has now become a part of our everyday lives. As children, many of us imagined the future through shows where robots did human tasks, heroes predicted events, or machines answered any question. These ideas, once seen as fantasies, are now real technologies shaping how we live. Fiction has turned into reality, with AI present in almost every field. Today, no matter your age, gender, profession, or background, almost everyone has heard of AI and experienced it in some way through digital assistants, recommendation systems, healthcare tools, workplace automation, and much more.

The fast growth of AI raises important questions: when the first practical systems appeared only a few decades ago, did anyone expect it to spread this quickly? Even with these advances, AI is still in an early stage, often needing human setup and supervision to work well. But the speed of progress makes us wonder: what will AI look like in 10, 20, or even 100 years? Could things once thought “impossible machines” equal to human intelligence, systems able to think in ways we cannot understand, or scenarios similar to time-travel or life-restoring stories, become possible? For now, AI is a powerful tool that still depends on us. Yet, will the time come when machines become smarter than people, with thought processes beyond our full understanding?

The global market for artificial intelligence is expanding at an extraordinary pace. According to Statista [1], AI technologies were valued at approximately 244 billion U.S. dollars in 2025, with forecasts suggesting this figure will surpass 800 billion by 2030. A separate analysis by Forbes [2] in March 2025 indicates that the market for AI tools and services alone grew by 31% in the past year and is expected to reach one trillion dollars by 2031. Beyond financial growth, AI adoption has become a daily reality for much of the world’s population. A large-scale study conducted by the University of Melbourne in partnership with KPMG (Trust, Attitudes and Use of Artificial Intelligence: A Global Study 2025) surveyed more than 48,000 people across 47 nations and revealed that two-thirds of individuals, around 66%, regularly engage with AI technologies [3]. Echoing this, Forbes (2025) projects that 378 million people will be using AI tools in 2025, a sharp rise from 116 million just five years earlier [2]. Remarkably, 64 million of these users were added in the last year alone, marking the largest year-on-year increase so far [2].

Organizations are also rapidly integrating AI into their operations. The Stanford HAI 2025 AI Index [4] reports that 78% of organizations currently use AI, up significantly from 55% in the previous year. The healthcare sector demonstrates one of the highest adoption levels [5], with 90% of hospitals now relying on AI systems for diagnosis and patient monitoring. Education is another area undergoing a transformation: 92% of students report using generative AI, compared to 66% in 2024 [6], with nearly one in five acknowledging that they have submitted AI-generated work. The business sector reflects a similar momentum; a Salesforce survey shows that 51% of marketers already employ generative AI, while an additional 22% intend to adopt it soon [7].

All the above statistics clearly demonstrate the profound integration of AI into society, underscoring both its rapid adoption and exponential growth. However, numerical measures of usage and market expansion only tell part of the story. To fully understand AI’s trajectory, it is essential to examine technology acceptance: a complex and multifaceted phenomenon shaped by user attitudes, organizational readiness, and societal perceptions. Foundational theoretical frameworks, most notably the Technology Acceptance Model (TAM) [8], and the Unified Theory of Acceptance and Use of Technology (UTAUT) [9], established that perceived usefulness, perceived ease of use, and social influence are core determinants of individuals’ intention to adopt new technologies, and these constructs have since been widely validated in AI-specific contexts [10,11].

Therefore, this raises fundamental questions:

• What does it mean to accept a technology?
• What factors drive individuals and organizations to embrace AI?
• Who is most inclined to do so?

Technology acceptance can be broadly defined as the willingness and readiness of individuals to adopt and use new technological systems, including artificial intelligence. It is closely tied to behavioral intentions, which are themselves influenced by perceptions of a technology’s usefulness and ease of use [12]. Acceptance is not merely an abstract concept; it directly determines how effectively AI can be implemented across domains such as education, healthcare, and the workplace [11]. Critically, trust has emerged in the literature as a key mediator of AI acceptance: unlike trust in human agents, trust in AI is constructed through perceptions of competence, transparency, and algorithmic explainability, and has been shown to be strongly domain-specific—individuals extend greater trust to AI in low-stakes contexts such as entertainment and education than in high-stakes domains such as healthcare and finance, where concerns about irreversible errors and moral accountability are more salient [13,14].

Additionally, multiple factors influence people’s willingness to engage with AI. Central among these are the perceived benefits: when individuals believe AI can enhance performance, productivity, or learning outcomes, they are far more likely to adopt it. For example, students who view AI tools as both beneficial to academic achievement and easy to operate demonstrate greater acceptance [10]. Attitudes toward technology also play a pivotal role; those with positive dispositions toward innovation consistently show higher levels of adoption [15]. Moreover, perceptions of control matter: individuals who feel empowered, with fewer barriers to adoption, are more open to embracing AI [16]. At the same time, risk perception plays an equally important role: research has identified privacy threats, algorithmic bias, job displacement, misinformation, and cognitive overreliance as the most salient AI-related risks in public perception [17,18], and these perceived risks do not simply reduce acceptance but rather increase public demand for governance and institutional safeguards [19].

On the other hand, social and cultural contexts further shape acceptance. In organizational settings, dimensions such as power distance, how strongly individuals are influenced by colleagues’ opinions, can be decisive. Acceptance is significantly higher when users perceive collective support and positive peer endorsement [20]. In educational contexts, familiarity with AI and perceptions of its benefits and challenges strongly influence adoption, particularly regarding generative AI tools like ChatGPT, which have reshaped students’ attitudes toward learning technologies [21]. At the same time, not all groups adopt AI equally. Those with greater technological readiness, such as employees in organizations actively preparing for AI integration, are typically more receptive [22]. Similarly, individuals who combine positive attitudes toward technology with a belief in AI’s ease of use and usefulness tend to adopt more readily [10]. Psychological dimensions also matter: trust in AI’s recommendations can strongly predict acceptance, and interestingly, broader worldviews, including religious or spiritual perspectives [23].

The widespread availability of AI, coupled with the willingness of individuals, organizations, institutions, and governments to adopt it in both short and long-term visions, highlights the need for a deeper understanding of public perspectives on AI. As it continues to evolve across political, scientific, and social domains, investigating public perception becomes not only timely but essential. A growing body of scholarship further underscores that AI governance, encompassing transparency requirements, accountability mechanisms, and regulatory oversight, is increasingly viewed by the public not as a barrier to innovation but as a prerequisite for sustainable and trustworthy AI adoption [24] This exploration allows for a more comprehensive view of how different populations engage with, interpret, and respond to AI. Both researchers and everyday users are increasingly involved in shaping Artificial Intelligence discourse, opening new avenues for dialogue and reflection.

Prior research on public AI perception has advanced considerably, with studies documenting adoption patterns, attitudinal drivers, and domain-specific trust across various populations [25]. However, several important limitations persist in the existing literature. First, the majority of studies rely exclusively on quantitative survey instruments, which capture measurable trends but are limited in their ability to uncover the reasoning, values, and emotional dimensions that underlie observed attitudes. Second, existing research tends to examine AI acceptance within single domains, such as healthcare [26] or education [27] in isolation, rather than exploring how individuals simultaneously negotiate trust and risk across multiple areas of daily life. Third, the geographic scope of published studies remains heavily concentrated in Western [28] and East Asian contexts [29], leaving populations in the Arab world and Middle East significantly underrepresented in the global evidence base on AI perception. Fourth, few studies have systematically examined the conditions under which individuals move from general awareness to conditional acceptance, that is, the specific requirements, such as privacy guarantees, transparency, and human oversight, that people impose before they are willing to engage with AI. These gaps collectively point to the need for empirically grounded, mixed-method research that captures both quantitative patterns and qualitative depth, across varied professional and cultural backgrounds, and with explicit attention to the negotiated and conditional nature of AI acceptance. A key concept guiding this study is that of conditional acceptance, the pattern whereby individuals engage with AI for its practical benefits while simultaneously imposing specific requirements, such as privacy guarantees, transparency, and human oversight, as preconditions for trust. This concept moves beyond binary acceptance/rejection frameworks and captures the negotiated, context-sensitive nature of public AI adoption.

Hence, against this backdrop, the present study is motivated by the recognition that public perception plays a pivotal role in the responsible and sustainable integration of AI into society. By combining quantitative methods with qualitative thematic analysis, this research provides a hybrid approach that captures both measurable trends and nuanced insights into public attitudes. This dual methodology represents a key advantage of the study, as it ensures a balanced understanding that goes beyond statistics to uncover the values, concerns, and expectations underpinning AI acceptance. The objectives of this study are threefold. First, examine how people from different backgrounds perceive and interact with AI technologies in their daily lives. Second, identify the main factors, such as trust, ethical considerations, and social influence, that shape acceptance or resistance. Third, analyze the diversity of perceptions across various backgrounds contexts, thereby providing a holistic picture of AI’s societal reception. The ultimate goal of this research is to inform policymakers, educators, and industry leaders about the drivers and barriers of AI acceptance, enabling them to design strategies that foster trust, transparency, and inclusivity. By offering a comprehensive perspective, this study not only contributes to academic discourse but also provides practical insights to guide the ethical and effective adoption of AI in different sectors.

Although the TAM and related frameworks provide important theoretical perspectives for understanding technology adoption, the present study was not designed to test a specific acceptance model. Instead, an exploratory mixed-method approach was adopted to capture a broader range of perceptions, trust considerations, risks, and governance expectations associated with AI across different participant groups. The central organizing concept guiding the analysis is conditional acceptance, the pattern whereby individuals engage with AI for its practical benefits while simultaneously imposing specific prerequisites for trust, which serves as both the analytical lens and the primary theoretical contribution of the study.

2. Study Design and Methodology

The main objective of this study is to obtain a comprehensive understanding that encompasses not only numerical patterns but also the underlying concepts, emotions, and concerns associated with artificial intelligence. To achieve this, a mixed-method research design was developed, combining quantitative and qualitative components to ensure both depth and breadth in the analysis. The research began with the development and distribution of an online survey aimed at reaching a wide range of participants. The instrument intentionally included open-ended questions in addition to multiple-choice items, allowing respondents to express their thoughts, concerns, and perceived advantages of AI in their own words. This qualitative element was essential, as it provided valuable insight into the emotional and cognitive dimensions of public attitudes and factors that cannot be adequately captured through numerical data alone.

Consistent with a convergent parallel mixed-method design, quantitative and qualitative findings are presented separately in the Section 4 and systematically integrated in the Discussion, where each data source is used to explain, confirm, or nuance the patterns identified by the other.

In parallel, the study examined the terms and phrases frequently used by participants when discussing AI. These linguistic patterns helped identify dominant narratives, concerns, and expectations related to technology adoption, offering a distinctive perspective on how individuals conceptualize AI and its potential impact on their daily lives. Hence, following the survey phase, several semi-structured interviews were conducted to further explore themes that emerged from the preliminary responses. These discussions provided deeper understanding of participants’ reasoning and experiences, particularly concerning perceived control, ethical considerations, and trust in human–AI interactions. Consequently, the methodological approach implemented in this research underscores the importance of integrating quantitative and qualitative research to create a balanced and comprehensive framework for examining public perceptions and acceptance of artificial intelligence.

2.1. Quantitative Phase: Survey Study

To investigate public perceptions of Artificial Intelligence, a structured online survey was designed and distributed to a heterogeneous sample of participants. A total of 115 individuals completed the survey, which was available in both English and Arabic to ensure inclusivity and accessibility across different linguistic and cultural backgrounds. The questionnaire aimed to capture demographic information, general awareness of AI, trust levels, perceived benefits and risks, and personal reflections on AI’s societal implications. It combined multiple-choice and open-ended questions to balance quantitative measurement with qualitative insights, allowing respondents to express their views in their own words when appropriate. The survey comprised 22 questions covering demographics, exposure to AI, usage patterns, perceptions, ethical considerations, and future expectations, where all questions were developed to provide a holistic understanding of how people from different backgrounds interact with and interpret AI technologies.

Table 1. Survey Questions and Response Options.

Q#	Question	Response Options
Q1	What is your age group?	Under 18; 18–24; 25–34; 35–44; 45–54; 55–64; 65+
Q2	What is your gender?	Male; Female
Q3	What is your highest level of education?	Primary/Elementary; High school; Bachelor’s; Master’s; Doctorate/PhD; Other
Q4	Which region/country do you live in?	Europe; USA; Canada; Latin America & Caribbean; Middle East/Arab countries; Africa; Asia (excluding Middle East); Oceania; Other
Q5	What is your current occupation/field?	Student; Academic/Researcher; Healthcare; Engineering/Technology; Business/Finance; Education/Teaching; Arts/Media; Government/Public sector; Freelance/Entrepreneur; Law; Non-profit; Other
Q6	When did you first hear the term “Artificial Intelligence”?	Before 2000; 2000–2010; 2011–2020; After 2020; After 2024
Q7	Where did you first hear about AI?	School/University; Workplace; News/Articles; Movies/TV; Social media; Friends/Family; Other
Q8	How would you describe your very first impression when you heard about AI?	Open-ended
Q9	How often do you use AI-powered tools today?	Daily; Several times a week; Several times a month; Rarely; Never
Q10	Which AI tools do you use most often?	Virtual assistants (Siri, Alexa, Google Assistant); Chatbots/LLMs (ChatGPT, Gemini, Copilot); Recommendation systems (Netflix, Spotify, Amazon, YouTube); Translation tools (Google Translate, DeepL); Productivity tools (MS Copilot, Grammarly, Excel AI); Image/Video tools; AI in gaming/entertainment; Other
Q11	What is the main reason you use AI tools?	Save time; Improve accuracy; Entertainment; Learning/Education; Work productivity; I do not use AI tools; Other
Q12	How would you describe AI’s impact on your daily life?	Very positive; Somewhat positive; Neutral; Somewhat negative; Very negative
Q13	In which area do you trust AI the most?	Healthcare; Education; Finance; Transportation; Entertainment; None (I do not trust AI); Other
Q14	What concerns you most about AI?	Job loss/automation; Privacy and data misuse; Bias and fairness; Misinformation; Security risks; Lack of transparency; Overreliance (“brain rot”)
Q15	Do you feel you fully understand how AI works?	Yes, very well; Somewhat; Basic level only; Not at all
Q16	Do you believe AI will have a positive or negative impact on society in the next 10 years?	Mostly positive; Mostly negative; Unsure
Q17	Do you support stronger regulations on AI development and use?	Yes, absolutely; Yes, but only in certain areas; No (may slow innovation); Not sure
Q18	In your opinion, what is the biggest opportunity AI will bring to humanity?	Open-ended
Q19	What is your biggest fear about AI?	Open-ended
Q20	Do you wish the world had never created AI?	Yes; No; Not sure
Q21	What do you believe will be the next major breakthrough after AI?	Open-ended
Q22	What message would you like to send to scientists and developers building the future of AI?	Open-ended

summarizes all survey questions and corresponding response options.

The selection of survey questions and response options was carefully structured to ensure comprehensive coverage of both demographic and perceptual dimensions related to Artificial Intelligence as displayed in

Table 2. Survey Question Categories and Analytical Focus.

Category	Questions	Analysis Focus
Demographics	Q1–Q5	Participant segmentation
Awareness & Exposure	Q6–Q8	First contact with AI
Usage	Q9–Q11	Behavior and motivation
Perception & Understanding	Q12, Q15, Q16	Attitudes and literacy
Trust & Regulation	Q13, Q14, Q17	Risk and governance views
Ethical & Emotional Insights	Q18–Q22	Hopes, fears, advice, reflection

. The first section (Q1–Q5) focused on demographic characteristics, including age, gender, educational attainment, geographic region, and occupational field. Collecting this information allowed for the classification of respondents into distinct social and professional segments, thereby enabling comparative analyses across different groups. Such demographic segmentation is crucial for identifying whether variations in awareness, trust, or ethical outlook toward AI correspond to differences in education, location, or professional exposure. Building upon this foundation, the second section (Q6–Q8) explored participants’ awareness and exposure to AI. Questions in this section examined when respondents first encountered the concept of AI, through which channels (e.g., school, media, or social networks), and their initial impressions. Linking this section to the demographic data made it possible to contextualize early exposure within broader socio-educational patterns, for instance, assessing whether younger or more technologically oriented participants tend to have earlier or more positive first impressions of AI.

The third section (Q9–Q11) transitioned from awareness to actual interaction, focusing on respondents’ usage patterns. It captured how frequently individuals engage with AI-powered tools, the types of applications they use, and their primary motivations for doing so. These variables provided behavioral evidence of how AI has been integrated into daily routines and professional workflows. When analyzed alongside the previous sections, these responses helped illuminate the relationship between familiarity and trust, whether frequent users perceive AI more favorably than those with limited exposure. Additionally, the fourth section (Q12, Q15, Q16) delved deeper into perception and understanding, assessing both the cognitive and affective dimensions of respondents’ attitudes. Participants evaluated AI’s impact on their personal lives and society, as well as their self-assessed understanding of how AI systems function. This section bridged the gap between behavioral interaction and ethical evaluation, revealing how perceived competence and literacy influence confidence and acceptance of AI technologies.

Extending this exploration, the fifth section (Q13, Q14, Q17) examined trust and regulation. Here, respondents identified the domains in which they felt most or least confident entrusting AI, expressed their principal concerns, such as privacy, fairness, or security, and shared their views on the need for regulatory oversight. By integrating these findings with the perception and usage data, the analysis could map the interplay between practical engagement, ethical apprehension, and the desire for institutional safeguards. Finally, the sixth section (Q18–Q22) consisted of open-ended, reflective questions designed to elicit deeper insights into participants’ moral reasoning and emotional stance toward AI. These items invited respondents to articulate in their own words the opportunities they foresee, their greatest fears, their visions for the future of technology, and the advice they would give to scientists and developers. When qualitatively analyzed and triangulated with earlier quantitative findings, this section enriched the interpretation by uncovering latent themes, values, and contradictions that structured questions might have overlooked.

2.2. Qualitative Phase: Interview Study

To complement the quantitative findings from the survey, a series of semi-structured interviews was conducted with selected participants representing different age groups, professional fields, and levels of familiarity with Artificial Intelligence. The purpose of these interviews was to gain deeper insight into how individuals conceptualize AI, perceive its presence in their daily lives, and articulate their emotions, values, and expectations beyond what can be captured through closed-ended survey questions. Each interview lasted approximately 20 to 30 min and followed an open-ended format that encouraged participants to express their thoughts freely while maintaining a consistent thematic structure across sessions. The guiding questions were as follows:

• I1: “If your friend or child asked you What is AI?, how would you explain it in simple words?”
• I2: “Can you think of a time you were using an app or a tool, and later found out it was actually using AI?”
• I3: “Imagine tomorrow all AI tools disappeared, what would be the hardest thing for you to live without?”
• I4: “Do you feel more excited or more worried about AI? Why? What do you hope AI will bring for the future?”
• I5: “Is there something you would never trust AI to do?”
• I6: “If you could ask AI to solve one big problem in the world, what would you choose?”
• I7: “If you could give one piece of advice to the people creating AI, what would you tell them?”

The choice of these interview questions was deliberate and grounded in the objective of eliciting personal, reflective, and experiential insights about AI that go beyond statistical representation. The first two questions were designed to explore Conceptual understanding and real-life exposure to AI, helping to identify how people define and recognize AI in everyday contexts, while questions three and four examined the emotional and cognitive dimensions of AI perception, probing excitement, fear, and hope to capture how individuals negotiate the duality of fascination and concern. On the other hand, question 5 (I5) targeted trust boundaries, revealing moral or psychological limits regarding AI’s role in decision-making, and finally, I6 and I7 encouraged forward looking and ethical reflections, offering a glimpse into participants’ visions for AI’s role in addressing global challenges and shaping future innovation.

These interview questions complemented the survey by providing narrative depth and emotional context to the numerical trends identified earlier. While the survey quantified levels of awareness, trust, and usage, the interviews illuminated the reasoning behind those attitudes, how people make sense of AI, the values they associate with it, and the boundaries they draw between human and machine roles. Together, the two methods formed a coherent and balanced framework for understanding public perception of Artificial Intelligence from both analytical and human, centered perspectives.

3. Participants and Data Collection

During this research, ethical considerations were prioritized to ensure participants’ privacy and comfort, encourage open and honest participation, minimize potential bias, and comply with ethical standards for studies involving human subjects. Hence, all survey responses were collected anonymously, with no personally identifiable information requested or stored at any stage. Similarly, for the qualitative interviews, participants’ names were intentionally coded to preserve confidentiality.

3.1. Survey Distribution

The quantitative component of the study was administered through an online survey created using Google Forms, and was distributed broadly across multiple platforms, including LinkedIn, Instagram, and Facebook, as well as through professional and personal WhatsApp groups. The primary aim of this open distribution strategy was to maximize diversity among respondents, ensuring representation from various age groups, nationalities, educational backgrounds, and professional sectors, and to allow this research to capture a comprehensive cross-section of public perceptions, thereby enhancing the external validity and representativeness of the findings. A total of 115 participants completed the survey, reflecting a heterogeneous sample that aligns with the study’s objective of investigating attitudes and perceptions toward Artificial Intelligence across multiple backgrounds. Figure 1 provides a visual overview of this participant distribution by age, education, and occupation.

A total of 115 participants completed the survey, comprising 71 females (61.7%) and 44 males (38.3%). The majority of respondents were between 35–44 years old (n = 36, 31.3%), followed by 18–24 years (n = 28, 24.3%) and 25–34 years (n = 25, 21.7%), with smaller proportions representing older age groups, 45–54 (n = 13, 11.3%), 55–64 (n = 8, 7.0%), 65 and above (n = 4, 3.5%), and under 18 (n = 1, 0.9%), as shown in Figure 1. In terms of geographical distribution, the sample was predominantly from the Middle East and Arab countries (74.8%), while others represented regions such as Europe, North America, and Africa, reflecting an international but regionally concentrated sample.

Regarding educational background, nearly half of the participants (46.1%) held a Bachelor’s degree, followed by Master’s degree holders (21.7%), high school graduates (15.7%), and a smaller group with doctoral degrees (11.3%) as depicted in Figure 2. Finally, in terms of professional background, respondents represented a variety of fields, with the highest proportions coming from education and teaching (18.3%), students (15.7%), and engineering/technology (14.8%). Other notable sectors included business and finance (12.2%), academic research (8.7%), freelance and entrepreneurship (7.8%), and healthcare (5.2%), alongside smaller contributions from government, media, and other professions.

3.2. Interview Sessions

The qualitative component of the study consisted of a series of interview sessions conducted individually or in small groups, designed to elicit reflective and experiential insights into participants’ understanding, emotions, and expectations regarding Artificial Intelligence in both daily life and professional contexts. In total, the interviews involved participants from a broad range of disciplines, career stages, and societal roles, including university students, academic staff and researchers, system specialists, engineers, managers, educators, finance and procurement professionals, healthcare-adjacent roles, and family groups combining parental and youth perspectives. This heterogeneity ensured that viewpoints extended from highly technical and professionally embedded AI use to informal, everyday interactions shaped by consumer platforms such as navigation systems, recommendation engines, and generative AI tools.

Table 3. Overview of Interview Sessions.

No.	Interview Code	Participants/Profile Description
1	INT_PE	Physical Education Coach (UK and Middle East)
2	INT_SA	Two Professors of Engineering/Academic Researchers
3	INT_AS	Sophomore Software Engineering Student
4	INT_SS	Academic Registration and Systems Specialist
5	INT_FA	Family group composed of parents and adult children
6	INT_DM	Professor of Management Information Systems (MIS)
7	INT_FI	Corporate Manager/Parent (Commercial and Cost Control Background)
8	INT_MA	Professional/Technical Advisor
9	INT_NA	General Participant (Non-academic, non-technical background)
10	INT_GM	General Manager
11	INT_AC	Accountant
12	INT_AR	Group of Architecture and Design Professionals
13	INT_ST	Statistician/Data-oriented Professional
14	INT_NS	Two Network Engineering Students

summarizes the interviewed groups.

Participants also demonstrated varying levels of experience and familiarity with AI, ranging from frequent, task-oriented users employing AI for professional problem-solving, system analysis, learning support, and productivity enhancement, to occasional or passive users whose exposure was largely indirect or incidental. Several interviewees occupied hybrid roles (e.g., professionals who are also parents or educators), allowing the study to capture layered perspectives that bridge workplace efficiency, ethical responsibility, education, and societal impact.

Additionally, semi-structured interviews were conducted face to face and audio-recorded with participants’ informed consent to ensure transcription accuracy and preserve the authenticity of spoken expressions. Audio recording enabled the capture of vocal nuances such as tone, emphasis, and spontaneous reflections, which are often lost in manual note-taking and written summaries. The recorded interviews were subsequently transcribed using TurboScribe AI, version 2.1, with careful verification to ensure fidelity to the original speech.

All interview recordings were first transcribed using TurboScribe AI. Following transcription, ChatGPT (version 5.2) was used solely as an organizational support tool to assist in structuring the interview transcripts by grouping participant responses according to the corresponding interview questions (I1–I7). ChatGPT was not used for coding, theme generation, interpretation, or analytical decision-making. The qualitative analysis was conducted using MAXQDA version 26.2. The transcripts were imported into MAXQDA, where responses were reviewed, organized, and analyzed using thematic analysis. The software facilitated data management, retrieval of relevant text segments, categorization of participant responses, identification of recurring patterns, and comparison across interview groups. Themes were identified through an iterative review of the transcribed data to uncover common perceptions, concerns, expectations, trust boundaries, and conditions influencing AI acceptance. All thematic decisions, interpretation of findings, and final analytical conclusions were performed by the researcher. Throughout the analysis process, the researcher repeatedly reviewed the original transcripts to ensure that the identified themes accurately reflected participants’ perspectives and the context of their responses.

4. Results

4.1. AI Awareness and Exposure

Understanding when and how individuals first became aware of Artificial Intelligence provides valuable insight into the social diffusion of emerging technologies and the factors shaping early perceptions. As shown in Figure 3, the timing of AI awareness reveals a clear generational gradient. A substantial proportion of participants aged 18–24 years reported that they first heard about AI after 2020, reflecting the growing prominence of AI applications in education, entertainment, and everyday life during the post-pandemic digital acceleration. Participants within the 25–34 and 35–44 age groups displayed a more distributed pattern of awareness, with many indicating initial exposure between 2011–2020, a period marked by the widespread adoption of intelligent systems such as voice assistants, facial recognition, and personalized recommendation platforms.

A similar pattern was observed among the youngest participants (under 18) and the oldest group (45 years and above), both of whom were more likely to report becoming aware of AI only after 2020. For these two groups, awareness appears to have emerged primarily in response to the recent public attention and global diffusion of generative AI technologies, reflecting exposure that is either very recent (for youth) or newly re-engaged (for older adults). This age-linked variation highlights how AI awareness develops in parallel with both technological access and life-stage exposure. Younger individuals are typically introduced to AI through informal, user-facing technologies, often within their digital learning environments, while mid-aged participants encounter it more directly in professional or applied contexts. The 35–44 age group appears to bridge these two spheres, combining early awareness with practical integration in work and education. These results are consistent with the demographic trends discussed earlier, where mid-career professionals demonstrated the most diverse and sustained engagement with AI, reinforcing the link between occupational diversity and technological literacy.

On top of this, the finding pattern was also reflected in how participants first encountered AI. Nearly half of the respondents (approximately 46%) reported learning about AI through social media platforms, followed by 21% through academic environments, 18% via friends or family, and about 11% in workplace or professional contexts. Hence, public awareness of AI is largely shaped by informal, digitally mediated exposure rather than formal instruction, particularly among younger participants who frequently engage with technology content online. At the same time, academic and professional settings continue to play a critical role in supporting more structured and sustained understanding, particularly among respondents in technical and educational fields. Beyond when and how participants first became aware of AI, their initial impressions reveal how emotional and cognitive responses accompany early exposure to emerging technologies. The open-ended responses to Q8 captured a spectrum of attitudes ranging from admiration and curiosity to fear and uncertainty. These responses were thematically analyzed and summarized in

Table 4. Thematic Analysis of Participants’ First Impressions of Artificial Intelligence (Q8).

Theme	Approx. Share	Illustrative Quotes
Amazement and admiration	∼40%	Impressive; Amazing; It can change everything
Curiosity and intellectual interest	∼22%	Science fiction becoming reality; eager to learn
Optimism and perceived progress	∼18%	Excited; making life easier; new era
Fear and moral concern	∼15%	Scary; replacing humans; double-edged sword
Neutral or uncertain	∼5%	Neutral; strange; hard to understand

.

4.2. AI Usage Patterns

Understanding how individuals use Artificial Intelligence tools in their daily lives provides an essential bridge between awareness and perception. While earlier section explored when participants first encountered AI and how they initially felt about it, this section examines the practical dimension of engagement, how often people use AI systems (Q9), which tools they rely on most frequently (Q10), and their key motivations for doing so (Q11). Consequently, a considerable proportion of participants reported using AI-powered tools either daily or several times a week, indicating that for many individuals, AI has transitioned from an occasional resource to a consistent part of digital interaction. Conversely, only a small minority indicated rare or non-existent usage, suggesting that disengagement from AI tools is increasingly uncommon in modern digital environments.

When examining the types of AI tools used (Q10), participants most frequently selected chatbots and language models such as ChatGPT, Gemini, and Microsoft Copilot as shown in Figure 4.

These systems have become popular due to their accessibility and broad functionality, supporting tasks like writing, summarization, translation, and problem-solving. Following these were virtual assistants (e.g., Siri, Alexa, Google Assistant) and recommendation systems embedded in platforms like Netflix, Spotify, YouTube, or Amazon. Notably, the tools used by daily users tended to be more interactive and generative, whereas those used by occasional or rare users were often passive systems integrated into entertainment or consumer platforms. This distinction highlights an important behavioral gradient: higher engagement corresponds to more active and purposeful forms of AI usage, while lower engagement aligns with incidental or background experiences of AI.

In addition, participants’ motivations (Q11) further illuminate this pattern. The most commonly identified reasons for using AI tools were saving time, improving accuracy, and supporting learning or education. Productivity oriented motivations were especially prominent among daily and weekly users, many of whom relied on AI for professional tasks such as document preparation, data analysis, scheduling support, and technical problem-solving. In contrast, participants who reported infrequent use of AI tools tended to cite more casual motivations, such as entertainment, curiosity, or occasional translation needs, suggesting that motivation intensity is closely linked to usage frequency.

4.3. AI Impact and Understanding

A comprehensive analysis of perceived AI impact (Q12, Q16), and understanding (Q15) reveals clear patterns shaped by both age and occupation. Overall, most respondents viewed AI’s influence on their daily lives as somewhat or very positive as illustrated in Figure 5, with younger participants (18–24) reporting the strongest positivity due to frequent exposure to AI-enabled tools across education, communication, and entertainment. Mid-career adults (35–44) tended to describe the impact as somewhat positive, emphasizing practical benefits such as efficiency, accuracy, and convenience, while older respondents (45+) were more neutral, often citing limited usage or uncertainty about AI’s broader implications. These age-based trends were reflected across occupational categories and where:

• Technology, engineering, and academic professionals reported consistently positive impact due to regular interaction with AI systems;
• Healthcare and business workers showed moderate positivity, acknowledging improvements in workflow and data accessibility;
• Educators expressed mixed perceptions, balancing instructional support from AI with concerns regarding academic integrity and student overreliance.

Self-assessed understanding of AI followed a similar gradient. While most respondents described their understanding as basic or somewhat adequate, younger participants tended to report a functional, though non-technical, familiarity. In contrast, many individuals aged 35–44, despite extensive workplace exposure, rated their understanding as basic, suggesting that usage does not always translate into conceptual literacy. Older adults (45+) were the most likely to indicate not at all, highlighting a persistent digital confidence gap. Occupational differences further reinforced these trends:

• Engineering, technology, and academic researchers demonstrated the highest levels of understanding;
• Business, healthcare, and government employees tended toward moderate or basic comprehension;
• Students, depending on their major, ranged widely from confident understanding among computing disciplines to minimal literacy in non-technical fields.

Nevertheless, expectations for AI’s societal impact over the next decade (Q16) also aligned with these demographic patterns. Younger respondents expressed strong optimism, viewing AI as a catalyst for innovation and problem-solving, while individuals aged 25–44 showed cautious optimism, acknowledging benefits but raising concerns about issues such as privacy, misinformation, and job displacement. Older participants leaned toward uncertainty, reflecting lower familiarity and perceived personal benefit. This occupational gradient was equally evident:

• Engineering, technology, academia, and business/entrepreneurship roles were the most optimistic, given their exposure to AI’s tangible benefits;
• Education and public-sector roles expressed higher uncertainty, frequently referencing ethical, fairness, and regulatory concerns.

4.4. AI Trust, Perceived Risks, and Regulation

Trust in Artificial Intelligence was strongly domain dependent. When asked “In which area do you trust AI the most?” (Q13), almost half of the respondents (47.8%, $n=55$) selected education, confirming the central role of AI as a learning and productivity assistant. Entertainment came second (14.8%, $n=17$), followed by an equally large group who chose “None–I do not trust AI” (14.8%, $n=17$). Healthcare attracted 7% of respondents ($n=8$), transportation 3.5% ($n=4$), and finance only 1.7% ($n=2$), while a heterogeneous “other/mixed” category (10.4%, $n=12$) included answers such as using AI mainly for language correction, fact checking, or technical information. The dominance of education is reflected in the professional breakdown: among teachers and students together, 62% selected education as the domain where they trust AI the most. This preference was further illuminated by interview data, where participants consistently framed AI as a learning assistant and productivity support tool—a role perceived as low-stakes and reversible, in contrast to the irreversible consequences associated with healthcare or financial decisions (INT_DM; INT_AS; INT_FA). By contrast, trust in healthcare remained more fragmented; within the small group of healthcare professionals, only half identified healthcare as the domain of highest trust, with the remainder split between education, entertainment, and no trust at all.

On the other hand, perceived risks show that participants are not blindly optimistic about AI, even when they use it frequently. Q14 allowed multiple selections of worries about AI as illustrated in Figure 6. The most frequently cited concern was “brain rot”/“overreliance on AI”, chosen by 62 respondents (53.9%). Close behind were privacy and data misuse (56, 48.7%), job loss and automation (54, 47.0%), and the spread of misinformation (49, 42.6%). Structural risks such as security vulnerabilities (31, 27.0%), lack of transparency (23, 20.0%), and bias and fairness issues (13, 11.3%) were less frequent but still substantial. Only 4 participants (3.5%) indicated that they were not worried about AI at all, and just one respondent (0.9%) explicitly mentioned fear of AI “taking over humans”. This pattern suggests that everyday cognitive and social harms (overreliance, privacy, misinformation, and employment) are far more salient than science fiction style catastrophe scenarios.

On the other hand, linking trust domains with these concerns reveals a nuanced picture. Even among those who trust AI most in education, almost half expressed worry about brain rot and job loss (each 47.3%), more than half were concerned about privacy (54.5%), and over a third pointed to misinformation (36.4%). Respondents who trusted AI primarily in entertainment were even more sensitive to informational and cognitive risks: about 70.6% of them selected both brain rot and misinformation as key concerns. Those who declared no trust in AI still tended to articulate specific worries rather than general rejection: 76.5% of this group mentioned brain rot, 58.8% privacy, and 47.1% misinformation. In other words, trust in particular AI applications does not eliminate awareness of systemic risks; instead, participants distinguish between useful domains of deployment and the broader social and cognitive costs that AI may introduce.

In addition, attitudes toward governance and regulation were correspondingly strong. In Q17, 71 respondents (61.7%) answered “Yes, absolutely” when asked whether they support stronger regulation of AI development and use, and a further 28 (24.3%) supported regulation only in certain areas. Together, this means that 86% of the sample favors some form of tighter regulation. Only 7 participants (6.1%) believed that regulation can slow innovation and should be avoided, while 9 (7.8%) were unsure. The support for regulation was high between age groups: in the 35–44 category, 78% selected “Yes, absolutely”, and even among the youngest adults (18–24 years), almost half (46%) chose absolute support with an additional 32% preferring regulation in specific areas. Frequent users were, if anything, more pro–regulation: 56% of daily users and 76% of those using AI several times a week opted for “Yes, absolutely”, with another 21–28% favoring targeted regulation.

To further examine whether attitudes toward AI differed across demographic groups, inferential statistical analyses were conducted using Chi-square tests of independence, with Cramér’s V reported as the corresponding effect size measure, as shown in

Table 5. Age-group differences in AI attitudes.

Variable	${\mathit{\chi }}^2$	p	V
Q13 (Trust Domain)	23.99	0.004	0.186
Q17 (AI Regulation)	11.47	0.245	0.182

. The analysis revealed a statistically significant association between age group and the domain in which participants reported the highest level of trust in AI (Q13) (${\chi }^2=23.99$, $p=0.004$, $V=0.186$), indicating that trust preferences varied across age categories, although the strength of the association was small. In contrast, no statistically significant association was observed between age group and attitudes toward stronger AI regulation (Q17) (${\chi }^2=11.47$, $p=0.245$, $V=0.182$), suggesting relatively consistent support for AI governance across different age groups, with the near-zero association confirming the practical equivalence of regulation attitudes regardless of age. Therefore, while trust in AI appears to be influenced by demographic characteristics, support for regulation is broadly shared across the surveyed population.

4.5. Ethical, Emotional and Reflective Vision

Q18 to Q22 provided a deeper understanding of how participants emotionally interpret the rapid integration of Artificial Intelligence into society. These questions revealed not only expectations and imagined opportunities, but also personal fears, ethical reflections, and advice directed toward AI developers. Hence, these qualitative responses offer an essential counterpart to the quantitative findings, capturing the human narrative behind attitudes toward AI.

4.5.1. Opportunities and Perceived Benefits (Q18)

When asked about the greatest opportunity AI will bring to humanity, participants highlighted themes related to efficiency, productivity, and knowledge expansion. The word cloud displayed in the following Figure 7 generated from all responses demonstrates a strong emphasis on time-saving, accessibility of information, faster productivity, healthcare advancement, automation, and the acceleration of scientific research. Words such as “time,” “information,” “knowledge,” “assistance,” “speed,” “healthcare,” “education,” and “productivity” appeared with the highest frequency. This indicates that respondents largely view AI as a catalyst for improving daily tasks and accelerating complex problem-solving processes across medicine, research, and education.

The word cloud is presented solely as a supplementary visualization of frequently occurring terms within participants’ responses to Q18. It was not used as the basis for coding, theme development, or interpretation. The findings reported in this section are derived from thematic analysis and semantic review of participants’ responses, with the word cloud serving only as an illustrative summary of commonly expressed concepts. Additional semantic patterns showed that many participants believe AI will enable breakthroughs in healthcare, particularly in diagnosing and managing diseases, while others emphasized AI’s ability to expand human knowledge, increase efficiency at work, reduce routine tasks, and support scientific discovery. These associations suggest that, despite concerns, respondents maintain a fundamentally utilitarian vision of AI, focusing on its potential to enhance human capability and reduce time spent on repetitive or cognitively demanding tasks.

4.5.2. Fears and Ethical Concerns (Q19)

In contrast with the generally positive outlook expressed in Q18, Question 19 elicited significantly more apprehensive responses. Participants were asked to describe their biggest fear about AI, and the results were overwhelmingly negative in sentiment. Using a semantic and ontological clustering approach, responses were categorized into thematic clusters, each representing a recurring pattern of fear or ethical concern. As illustrated in Figure 8, the dominant concerns centered around job displacement, loss of critical thinking, and fears of AI surpassing human control. To further contextualize these insights,

Table 6. Thematic Clusters of AI-Related Fears Expressed by Participants (Q19).

Cluster	Approx. Share	Illustrative Topics
Job Loss and Replacement	∼33%	Job loss; routine jobs disappearing
Loss of Critical Thinking	∼29%	Brain rot; excessive dependency
Misinformation and Bias	∼11%	Fake information; manipulation
Loss of Control	∼17%	AI takeover; systems out of control
Privacy and Security	∼8%	Data leaks; surveillance
Ethical and Cultural Concerns	∼5%	Impact on values; children; spirituality

summarizes the six major clusters alongside representative topics extracted from participant responses. Overall, the distribution of sentiment was predominantly negative (approximately 87%), with only a small proportion of neutral (around 7%) or mixed/uncertain responses (6%), and no explicitly positive sentiments reported.

These clusters reveal a multifaceted ethical landscape in which participants articulate fears about economic, cognitive, social, and existential threats. Notably, many of these fears map directly onto the risks identified in Q14, confirming a strong alignment between structured and unstructured responses.

4.5.3. Reflections on the Future of Technology (Q21)

To explore perceptions about future technological frontiers, Question 21 asked participants to predict what major breakthrough might follow Artificial Intelligence. More than half of the respondents expressed uncertainty: 59 individuals (51.3%) stated that they did not know or were unsure, indicating that many participants view AI as such a transformative milestone that imagining what comes next remains difficult. A further 33 participants (28.7%) provided diverse and ungrouped answers that did not align with a single thematic category. Among the more defined technological expectations, 9 respondents (7.8%) believed that the next major breakthrough would be a more advanced or evolved form of AI itself. Biotechnology and medical innovation were mentioned by 5 participants (4.3%), reflecting expectations of progress in genetic engineering, disease eradication, and personalised medicine. Quantum computing appeared in 4 responses (3.5%), while another 4 participants (3.5%) anticipated forms of human–machine integration such as brain–computer interfaces or cyborg-like augmentation. Finally. only one respondent (0.9%) envisioned future advancements in space exploration.

4.5.4. Messages to Scientists and Developers (Q22)

In the final open-ended question (Q22), participants were invited to share a message with scientists and developers shaping the future of AI. Much like the concerns expressed in Q19, these responses were largely cautionary in tone. Many urged developers to prioritize ethical safeguards, transparency, and responsible innovation, emphasizing the need for stricter regulation, protection of user data, and fairness in algorithmic outcomes. For instance, several respondents highlighted the importance of “ensuring AI does not replace human judgment” and “preventing misuse that could harm society.” Others stressed the need to develop AI systems that enhance rather than replace human capabilities, underscoring the importance of preserving critical thinking, human dignity, and core social values. Representative comments included appeals such as “please don’t let AI weaken our thinking skills” and “design technology that supports, not controls, humans.”

A number of participants encouraged scientists to make AI accessible and beneficial to all, voicing concerns about corporate control or unequal access. Some urged developers to “keep AI fair and available to everyone,” while others warned against “allowing only large companies to dictate how AI evolves.” A smaller subset expressed optimism and appreciation for technological progress, acknowledging AI’s potential to advance healthcare, education, and environmental sustainability. These responses conveyed hope for “solutions to diseases,” “better learning tools,” and “technology that saves lives.”

4.6. Interview Results: Determinants of Public Attitudes Toward AI

This subsection reports the qualitative findings from the semi-structured interviews. Thematic analysis was conducted to identify recurrent determinants shaping participants’ attitudes toward AI. Across the fourteen interviews, participants represented diverse roles and backgrounds, including students, parents, academics, professionals, and technical staff, allowing for cross-role comparison and thematic triangulation. Overall, interviewees described AI as a highly useful assistant for learning and productivity, while simultaneously expressing strong concerns about privacy, misuse, overreliance, and future labor disruption. Notably, these perceptions were not mutually exclusive; many participants articulated both appreciation and apprehension, often within the same narrative. The qualitative findings are organized into five main categories: (1) concerns, (2) expectations, (3) environmental influences, (4) individual characteristics, and (5) minimum requirements for acceptable AI use.

4.6.1. Concerns

Participants repeatedly emphasized uncertainty regarding what happens to information shared with AI systems, especially in professional contexts. This uncertainty was expressed across both technically oriented and non-technical participants, suggesting that data-related anxiety is not limited to lack of expertise. However, some expressed skepticism that confidentiality can be guaranteed in practice:

“I don’t fully trust AI in general, mainly because of over-reliance and confidentiality issues.”

[INT_FI, Corporate/Parent]

Moreover, concerns were often framed in concrete scenarios such as uploading internal reports, financial data, or sensitive content into AI tools, particularly in corporate, academic, and administrative environments:

“Someone might upload confidential company data… Even if AI claims governance, I believe there are no real secrets anymore.”

[INT_FI, Corporate/Parent]

In addition to privacy-related risks, a second major theme was the fear that heavy dependence on AI could reduce independent thinking, especially among younger users. This concern was echoed by parents, educators, and family groups, who viewed cognitive offloading as a long-term developmental risk. This was often expressed through worries about replacing reasoning with instant answers:

“Teenagers, in particular, talk to AI constantly.”

[INT_DM, Academic]

Consequently, participants linked overreliance to long-term social and cognitive risks, including reduced critical thinking, weakened accountability, and emotional dependency:

“AI should not be expected to answer everything, and it should never replace verified sources.”

[INT_DM, Academic]

Similarly, interviewees frequently mentioned uncertainty about the future of work and the possibility of role replacement in specific sectors. These concerns were expressed by students anticipating career entry, as well as by parents and professionals reflecting on labor market shifts:

“I’ve heard many stories about people losing their jobs because of AI.”

[INT_AS, Student]

Furthermore, some participants perceived certain professions as particularly exposed to automation due to AI’s ability to process large bodies of text and precedent, with legal, administrative, and knowledge-intensive roles frequently cited:

“AI excels at analyzing large datasets and referencing similar cases.”

[INT_FI, Corporate/Parent]

Beyond professional implications, several interviewees were worried that AI could reduce genuine human connection and alter social behavior. This concern appeared strongly when participants discussed children and family dynamics, particularly the substitution of parental guidance or interpersonal dialogue with AI-mediated interaction:

“Children should ask their parents, not ask AI instead.”

[INT_FA, Family group]

Finally, others emphasized the risk that AI could shape identity, voice, or social expression in subtle ways, particularly through automated writing, messaging, or personalization systems:

“It can speak for you without sounding like you.”

[INT_MA, Professional]

4.6.2. Expectations

Despite concerns, interviews reflected strong expectations about AI’s benefits. These expectations were consistently expressed across all participant categories, including those who were otherwise cautious or worried. In particular, three expectation clusters dominated: (1) productivity and efficiency, (2) learning support and cognitive organization, and (3) problem-solving potential (especially in complex domains).

Consequently, AI was widely framed as a time-saving assistant that accelerates routine cognitive tasks such as summarization, idea generation, and organizing information. Participants across academic, professional, and technical roles emphasized time efficiency and task simplification as key advantages:

“AI is a system that learns, analyzes, and predicts, like having a smart partner.”

[INT_MA, Professional]

Additionally, participants often described AI as useful for structuring thoughts and improving clarity, especially in writing, studying, and conceptual organization. This function was highlighted by students, educators, and design-oriented participants alike:

“One important role of AI… is that it helps organize ideas.”

[INT_DM, Academic]

Finally, beyond individual-level benefits, some participants expressed optimistic, future-oriented expectations, imagining AI contributing to large-scale improvements such as disease treatment, educational accessibility, humanitarian coordination, and sustainability. However, this optimism was typically conditioned on responsible governance and careful boundaries, with participants repeatedly emphasizing that such benefits depend on ethical deployment rather than technological capability alone.

4.6.3. Environmental Influences

Environmental influences refer to external conditions that shape attitudes toward AI. In this context, three determinants were most common: (1) social media exposure and public narratives, (2) institutional norms (education/workplace), and (3) governance and accountability expectations. These influences operated as contextual amplifiers, shaping both enthusiasm and fear.

Firstly, participants noted that AI awareness spreads largely through social media and online discourse, where AI is often presented as either miraculous or threatening. Exposure to science-fiction narratives, viral news stories, and algorithm-driven platforms contributed to polarized perceptions. Consequently, this exposure shapes expectations and fear simultaneously.

Aside from public narratives, workplace context influenced both adoption and caution. Professionals and technical staff framed AI acceptance as dependent on organizational rules, data policies, and responsibility structures:

“Used correctly… AI is powerful and helpful. It just needs clear rules and accountability.”

[INT_MA, Professional]

Moreover, participants repeatedly highlighted the need for oversight and safeguards at institutional and governmental levels. Several interviewees emphasized that progress must be paired with control to prevent misuse or loss of human agency:

“For every new step you take, take three steps in control.”

[INT_MA, Professional]

4.6.4. Individual Characteristics

Attitudes varied depending on personal experience, profession, and technical familiarity. In this respect, two key characteristics emerged: (1) AI literacy/technical confidence, and (2) personal role (e.g., parent, student, professional), which shaped risk sensitivity. These characteristics influenced not only perceived benefits and risks but also trust thresholds. Hence, more technically oriented participants tended to express nuanced positions, namely strong recognition of AI’s usefulness paired with strong emphasis on data handling, security, and ethical design. Rather than unconditional trust, technical familiarity often resulted in more explicit boundary-setting.

On top of technical familiarity, role-based sensitivity played a significant role. Parents emphasized child development, emotional resilience, and social effects, while students highlighted job fears, educational fairness, and trust boundaries. Meanwhile, professionals emphasized confidentiality, responsible usage, and institutional accountability, reflecting their exposure to regulated environments.

4.6.5. Minimum Requirements for Trustworthy AI Use

Beyond general attitudes, interviewees proposed implicit “must-have” requirements for AI to be considered acceptable. These requirements were remarkably consistent across roles and levels of expertise. In particular, five minimum requirements were consistently reflected: (1) data privacy and confidentiality, (2) transparency and user awareness, (3) human oversight in high-stakes contexts, (4) ethical training data and fairness, and (5) clear usage boundaries.

First and foremost, the strongest requirement was that AI tools should not expose personal or institutional data. This concern was shared by students, parents, professionals, and technical staff alike:

“When I upload information… I always wonder: Who can see this?”

[INT_MA, Professional]

In addition to privacy concerns, participants stated clear trust boundaries, especially regarding medicine and finance, which were consistently identified as high-risk domains requiring human accountability:

“I wouldn’t trust AI with medicine.”

[INT_AS, Student]

“I would not trust AI with anything related to banks.”

[INT_NA, Participant]

On the other hand, participants emphasized safe data foundations and ethical safeguards, highlighting the responsibility of developers and institutions in shaping trustworthy systems:

“Make sure AI systems are built on secure, trusted, and ethical datasets.”

[INT_SA, Academic/Engineering]

4.7. Key Findings and Research Question Synthesis

The findings of this study provide several important insights into public perceptions and acceptance of Artificial Intelligence. First, AI has become deeply integrated into participants’ daily lives, with most respondents reporting regular interaction with AI-powered tools, particularly generative AI systems, virtual assistants, and recommendation platforms. Overall perceptions of AI were predominantly positive, reflecting its growing role in supporting productivity, learning, and access to information. Second, acceptance of AI was found to be conditional rather than unconditional. While participants recognized substantial benefits associated with AI adoption, they simultaneously expressed concerns regarding privacy, misinformation, overreliance, job displacement, transparency, and accountability. Trust was strongly domain-dependent, with education emerging as the most trusted application area, whereas healthcare and finance attracted lower levels of trust due to the perceived consequences of incorrect decisions. Third, both the survey and interview findings highlighted the importance of governance, ethical safeguards, and human oversight. Strong support for AI regulation suggests that participants do not view governance as a barrier to innovation, but rather as a necessary condition for responsible and trustworthy AI deployment. Consequently, public acceptance of AI appears to be shaped by a continuous evaluation of benefits, risks, trust, and accountability, supporting the concept of conditional acceptance proposed in this study.

5. Discussion

5.1. Socio-Demographic, Cognitive, and Behavioral Foundations of AI Acceptance

The demographic and professional composition of the sample (Q1–Q5) provides critical context for interpreting patterns of AI awareness, usage, and evaluation. The predominance of participants aged 18–44 situates the study within a cohort that has experienced both pre-AI and AI-saturated digital environments. This generational positioning explains the coexistence of normalization and disruption observed across responses. Younger participants, whose first exposure to AI frequently occurred after 2020 (Q6), tend to perceive AI as an integral, almost invisible layer of everyday technology, whereas older respondents more often frame AI as an accelerating force that challenges established cognitive, professional, and ethical frameworks. Educational attainment and occupational field further mediate acceptance: respondents with higher education levels and roles in education, engineering, research, and management demonstrate higher usage frequency (Q9), broader tool adoption (Q10), and more articulated perceptions of benefits and risks. This indicates that acceptance is not merely driven by access, but by interpretive capacity, the ability to contextualize AI within professional practice and social responsibility.

Patterns of awareness and early exposure (Q6–Q8) reveal that AI literacy is shaped largely outside formal education. Social media emerged as the primary channel through which participants first encountered AI, followed by academic settings and interpersonal networks. This fragmented exposure helps explain the emotional polarity of first impressions, where fascination and optimism dominate but are accompanied by fear, moral concern, and uncertainty. Interview responses to I1 and I2 reinforce this interpretation: participants typically defined AI in functional, outcome-oriented terms and often realized only retrospectively that tools they routinely used were AI-driven. This delayed recognition underscores AI’s embeddedness and invisibility, which lowers barriers to adoption while simultaneously fueling distrust once agency and data implications become salient.

Behavioral engagement patterns (Q9–Q11) reveal a clear gradient from incidental exposure to instrumental dependence. Daily and weekly users integrate AI deeply into learning, work productivity, and cognitive organization, favoring interactive tools such as chatbots and language models. Infrequent users, by contrast, engage primarily with passive AI systems embedded in entertainment and recommendation platforms. Interview responses to I3 confirm that AI has already become a form of cognitive infrastructure: participants consistently identified writing support, information synthesis, translation, and planning as the most difficult capabilities to relinquish in an AI-free scenario. This suggests that acceptance is reinforced by perceived indispensability rather than ideological alignment, with familiarity cultivating reliance and reshaping expectations about speed, efficiency, and intellectual effort.

5.2. Trust, Risk Perception, and Governance as Conditions for Acceptance

Despite widespread adoption, perceived understanding of AI (Q15) remains limited, even among frequent users. Most participants reported only basic or partial comprehension of how AI systems function, revealing a structural gap between use and literacy. Interview narratives consistently framed AI as a “helper” or “assistant,” rarely engaging with algorithmic mechanisms or data dependencies. This gap is analytically significant, as it increases vulnerability to overreliance and automation bias. The prominence of “brain rot” and loss of critical thinking as the most frequently cited concern (Q14) reflects an implicit awareness that convenience may undermine cognitive engagement, even when users cannot articulate the technical causes of this risk.

Trust in AI (Q13) emerges as strongly domain-specific and risk-weighted. Education and entertainment are perceived as safe domains due to their reversibility and low stakes, whereas finance and healthcare remain domains of pronounced hesitation. Interview responses to I5 align directly with this pattern, as participants articulated clear trust boundaries, rejecting AI autonomy in contexts involving irreversible consequences or moral accountability. Interview responses further explain why education attracts the highest trust: participants described AI in educational contexts as a helper that supports rather than replaces human judgment, where errors carry limited consequences and human oversight remains easily exercisable. By contrast, the low trust in healthcare and finance reflects participants’ perception that decisions in these domains are irreversible, morally consequential, and require accountability that AI cannot yet provide (INT_AS; INT_NA; INT_FI). Importantly, distrust does not manifest as rejection but as selective containment. Even respondents who selected “no trust” in any domain expressed conditional acceptance in specific low-risk use cases, indicating that skepticism operates through boundary setting rather than resistance.

Attitudes toward regulation (Q17) further clarify this dynamic. Strong support for regulation, endorsed by 86% of participants, demonstrates that governance is perceived as a prerequisite for sustainable adoption rather than a barrier to innovation. Notably, respondents with the lowest trust levels were among the strongest proponents of regulation, suggesting that skepticism amplifies demand for safeguards. Interview responses to I7 reinforce this interpretation, as participants framed regulation as necessary to ensure accountability, protect data, and preserve human agency. Governance, in this context, functions as a trust-enabling mechanism that allows users to reconcile AI’s benefits with its risks.

5.3. Ethical Reflection, Future Orientation, and Conditional Optimism

Expectations regarding AI’s societal impact (Q16) reveal cautious optimism tempered by uncertainty. While most participants anticipate a positive influence over the next decade, this optimism is explicitly conditional, often tied to ethical use, transparency, and regulatory oversight. The difficulty many respondents experienced in envisioning a technological breakthrough beyond AI (Q21) underscores the perceived magnitude of the current transformation. AI is not conceptualized as one innovation among many, but as a meta-technology reshaping how progress itself is imagined. Interview responses to I4 and I6 reinforce this perception, as participants expressed hope that AI could address global challenges such as healthcare access, education quality, and sustainability, while simultaneously warning against concentration of power and erosion of human judgment.

Fear-related responses (Q19) and reflections on a world without AI (Q20) deepen this ambivalence. Dominant fears center on job displacement, misinformation, loss of control, and cognitive dependency rather than catastrophic or science-fiction scenarios. This indicates that AI anxiety is pragmatic and anticipatory, grounded in concern for social stability and human capability. Responses to Q20, particularly expressions of uncertainty rather than outright rejection, reveal moral tension rather than regret. Participants appear unwilling to forgo AI’s benefits, yet uneasy about its long-term implications. Interview data amplify this tension, especially among parents and educators, who emphasized developmental, relational, and social dimensions of AI’s impact.

Across both survey and interview data, a consistent normative stance emerges: AI is valued as a means, not an end. Participants repeatedly emphasized that AI should enhance rather than replace human judgment, preserve dignity, and operate within clearly defined boundaries. Acceptance, therefore, is neither unconditional nor static, but negotiated. It evolves through continuous evaluation of utility, risk, governance, and social consequence. This negotiated acceptance challenges simplistic dichotomies of enthusiasm versus resistance and highlights the public’s capacity for reflective engagement with emerging technologies. The findings suggest that long-term societal acceptance of AI will depend less on increasing technical capability and more on embedding AI within frameworks of trust, accountability, and human-centered design.

6. Conclusions and Future Work

This study set out to examine public perception and acceptance of Artificial Intelligence through a mixed-method design that combines a structured survey ($N=115$) with semi-structured interviews across varied professional and social profiles. Overall, the findings indicate that AI has transitioned from a novel concept to an embedded socio-technical infrastructure shaping daily routines, learning practices, and workplace productivity. At the same time, acceptance is neither uniform nor unconditional. Participants’ responses demonstrate a consistent pattern of conditional adoption: individuals frequently engage with AI tools for practical benefits (e.g., efficiency, learning support, and information accessibility), while simultaneously maintaining concerns about cognitive overreliance, privacy, misinformation, and employment disruption. In particular, trust was strongly domain-dependent, with education emerging as the most trusted application area, whereas finance and healthcare remained domains of hesitation and stricter human oversight expectations.

Crucially, the results show that lower trust does not necessarily translate into resistance to AI; instead, it increases demand for governance and safeguards. The strong support for regulation observed across demographic groups suggests that public acceptance is closely tied to perceived accountability, transparency, and boundary-setting rather than to technological enthusiasm alone. The interview narratives reinforced these trends by revealing how participants conceptualize AI primarily as an assistant and productivity partner, while drawing firm trust boundaries in high-stakes contexts and emphasizing ethical responsibility, confidentiality, and the preservation of human judgment. Taken together, the study contributes empirical evidence that AI acceptance is best understood as a negotiated process shaped by demographic positioning, usage intensity, perceived literacy, domain-specific risk, and expectations of institutional control.

Future work should extend these findings in four main directions. First, the sample was regionally concentrated, and subsequent studies should increase cross-regional balance and sample size to enable statistically robust comparisons across cultures, regulatory environments, and socioeconomic contexts. Second, the survey instrument used in the present study was developed specifically for this exploratory investigation and included several single-item measures to capture broad perceptions while maintaining accessibility and completion rates. Although this approach provided useful descriptive insights, future survey iterations should incorporate validated multi-item scales from technology acceptance and trust frameworks (e.g., perceived usefulness, perceived ease of use, trust propensity, perceived risk, and social influence), and conduct psychometric validation procedures including reliability assessment (e.g., Cronbach’s alpha) and exploratory and confirmatory factor analysis. This would allow for predictive modeling through regression or structural equation modeling and enable stronger theoretical generalization across populations and cultural contexts. Third, the literacy gap identified between frequent AI usage and limited understanding of how AI works suggests the need to experimentally evaluate the effect of AI education interventions (e.g., short explainability modules, privacy literacy prompts, or critical-thinking scaffolds) on trust calibration, overreliance, and risk perception. Finally, longitudinal research is essential to capture how attitudes evolve as AI systems become more integrated into education, employment, and public services; tracking the same participants over time would clarify whether concerns such as “brain rot,” misinformation, and job displacement diminish through adaptation or intensify as dependence grows. Collectively, these directions would strengthen the external validity of public perception research and support the design of human-centered AI policies and educational strategies that foster informed, responsible, and sustainable adoption.