Research Methodology

Generative AI Usage Behavior, Literacy, and Perceived Risks

• This section explores Taiwan's population's experience with generative AI, examines current risk perception and literacy regarding generative AI, and further investigates attitudes toward legal regulation of AI technology and promotion of AI literacy education in contemporary society.

Current Status and Experience of Generative AI Usage

AI Literacy Among Taiwan's Population: Awareness Literacy

• AI literacy refers to users' ability to correctly identify, properly use, and clearly evaluate AI technology within ethical norms. This survey asked about agreement with: "I am capable of distinguishing which products and services use AI and which do not (e.g., AI customer service versus human customer service online)." Results show 70.19% tend to agree (including "strongly agree" and "agree," same hereinafter), meaning most internet users in Taiwan believe they can clearly distinguish between AI and non-AI.

AI Literacy Among Taiwan's Population: Usage Literacy

• For usage literacy, the survey asked: "I am capable of using AI products or services to accomplish what I want to do." Results show 66.23% tend to agree, meaning over 60% of internet users in Taiwan believe they can use AI products or services to meet specific needs.

AI Literacy Among Taiwan's Population: Evaluation Literacy

• For evaluation literacy, the survey asked: "After using AI products or services, I clearly know their advantages and disadvantages." Results show 79.25% tend to agree, with the vast majority of internet users in Taiwan believing they can clearly discern the strengths and weaknesses of AI products or services.

Risk Perception of AI Technology: Misinformation (Impact on Self)

• This section explores two types of generative AI risks—"misinformation" and "privacy breaches"—examining perceived impacts on "self" and "others," combined with framing effects theory from communication studies to test whether questioning tone affects respondents' risk assessment. Two frames were tested: "certainty framing" with definitive tone (e.g., "will produce bias"); "possibility framing" with probabilistic tone (e.g., "may produce bias"), observing how different communication approaches influence public risk perception.
• When asked using "certainty framing"—informing that generative AI "will contain bias or produce misinformation"—20.14% of internet users in Taiwan consider this risk to have greater impact on themselves. However, when asked using "possibility framing"—informing that generative AI "may contain bias or produce misinformation"—the proportion drops slightly to 17.02%.

Risk Perception of AI Technology: Misinformation (Impact on Others)

• When evaluating AI risk impact on "the general public" (others) using "certainty framing"—informing that generative AI "will contain bias or produce misinformation"—62.06% of internet users in Taiwan consider this risk to have greater impact on the general public. Using "possibility framing"—informing "may contain bias or produce misinformation"—the proportion is 60.68%.

Risk Perception of AI Technology: Privacy Breaches (Impact on Self)

• When asked using "certainty framing"—informing that private sensitive information sent when using generative AI "will be leaked and misused"—36.60% of internet users in Taiwan consider this risk to have greater impact on themselves. Using "possibility framing"—informing "may be leaked and misused"—the proportion increases to 39.47%.

Risk Perception of AI Technology: Privacy Breaches (Impact on Others)

• When asked using "certainty framing" about privacy "will be leaked and misused" impact on the general public (others), 60.43% of internet users in Taiwan consider the impact greater. Using "possibility framing" for "may be leaked and misused," the proportion is 62.01%.

Taiwan Public Opinion on Legal Regulation of Generative AI

• The survey asked respondents: "Do you support that generative AI products or services like ChatGPT should be subject to legal regulation?" Results show 82.74% tend to support (including "strongly support" and "support," same hereinafter); 10.99% tend to oppose (including "strongly oppose" and "oppose," same hereinafter), highlighting that mainstream Taiwan public desires relevant legal control of generative AI.

Taiwan Public Opinion on Promoting AI Literacy Education

• The survey asked respondents: "Do you support that when people use generative AI products or services like ChatGPT, they should receive AI literacy training or courses, such as improving AI usage skills, understanding AI risks, and critical thinking?" Results show 83.03% tend to support; 10.23% tend to oppose. This similarly shows over 80% of Taiwan's population believes users of generative AI products or services should receive relevant training and courses.

Online News Consumption Behavior

Traditional news media no longer appears to be the primary channel for people to obtain information, but merely one type of information people need. This survey explores Taiwan's news and information consumption behavior through dimensions including news sources, fact-checking capabilities, news avoidance attitudes, and public perception of Taiwan's online opinion climate.

News Sources

• Consistent with last year, in terms of "individual channels" for obtaining news, "television" from traditional news media remains the most common source among Taiwan's population at 32.02%, though this represents a significant decline from 39.57% in 2024.
• When analyzed from a platform perspective, "digital platforms" such as news portals, social media, instant messaging, and YouTube as primary news sources have consistently exceeded "traditional media" such as television and print newspapers.
• 

Source: 2025 Taiwan Internet Report, conducted from July 28 to September 1, 2025, weighted values. Sample size: 2,142 (dual-frame sampling, all samples).

News Avoidance: Is News Useful?

News avoidance encompasses behavioral active avoidance, algorithm-driven passive avoidance, and attitudinal avoidance among audiences. In the past three months, 10.04% of Taiwan's population frequently actively avoided watching news, while 7.73% sometimes actively avoided watching news. These two categories constitute news avoiders, totaling 17.77%. 12.57% rarely actively avoided watching news, and 69.17% never actively avoided watching news. These two categories constitute non-avoiders, totaling 81.74%. The proportion of news avoiders in Taiwan (17.77%) far exceeds those who don't watch news at all (2.13%), indicating that for Taiwan's population, actively avoiding news is more an expression of attitude toward news value rather than complete non-contact with news content.

Cognitive Warfare

• Cognitive warfare is an emerging form of conflict in contemporary global competition, integrating psychological warfare, information warfare, and narrative warfare. It aims to gradually shape or guide public understanding of policies, international situations, and friend-foe perceptions through disinformation, agenda manipulation, and opinion control. For Taiwan, failure to take this seriously and establish defenses could lead to manipulation of public discourse space, erosion of public trust, and long-term challenges to democratic institutions and national security.

Public Awareness of Influencer-Based Cognitive Warfare

• This survey asked: "Media reports suggest that 'foreign forces may influence Taiwan's public opinion through video creators (such as YouTubers and influencers).' How familiar are you with this matter?" Taiwan's population shows divided familiarity with "foreign forces potentially influencing Taiwan's public opinion through video creators (such as YouTubers and influencers)." Overall, 43.31% indicate they are familiar or very familiar; conversely, 51.91% consider themselves unfamiliar or very unfamiliar.

Online Resources for Countering Cognitive Warfare

• When facing external forces infiltrating public discourse through disinformation and emotional manipulation, the key to timely public identification and response lies in "accessible and usable public online resources." This survey asked: "Are there sufficient resources online to help you resist cognitive warfare attacks from foreign countries?" Results indicate Taiwan's population generally believes online resources for resisting foreign cognitive warfare attacks are insufficient. Nearly 60% (52.35%) of internet users selected "insufficient" or "very insufficient."

Information Verification Behavior Among Taiwan's Population

• In an environment where cognitive warfare spreads, information verification behavior is key to maintaining social resilience. This survey asked: "How often do you search online for materials (including articles or videos) teaching how to identify disinformation, use fact-checking tools, or discuss with others online to determine information credibility?" Results show that Taiwan's population's frequency of proactively using articles, videos, fact-checking tools, or discussing with others to determine information authenticity is generally low. Over 60% (65.91%) indicate they "never" or "rarely" do this; only 13.60% frequently or always engage in verification behavior.

Self-Efficacy in Identifying Politically Manipulated Messages

• This survey asked whether people "have the ability to identify cognitive warfare or messages with specific political purposes." Results show most of Taiwan's population holds relatively confident attitudes toward their ability to identify cognitive warfare or messages with specific political purposes. Nearly 73% (73.37%) of respondents indicated "agree" or "strongly agree."

Risk Perception of Cognitive Warfare

• This survey asked about the perceived severity of cognitive warfare attacks on Taiwan. Results show most of Taiwan's population believes Taiwan faces serious cognitive warfare attacks. Over 65% (65.86%) of respondents consider the situation "serious" or "very serious," reflecting that cognitive warfare is viewed as a widespread and highly threatening problem.

Public Trust in Government Crisis Management

• Survey results indicate Taiwan's population shows divided trust in whether the government can make correct decisions when facing crises. Approximately 42% (41.88%) of respondents indicated "agree" or "strongly agree." The proportion holding negative attitudes is higher at 45.02%, reflecting that nearly half the population harbors doubt or distrust toward government crisis response capabilities.

Public Perception of Taiwan's Social Resilience

• This survey asked whether people agree with the statement "society has handled crises well in the past and will do so in the future," finding clearly divided attitudes. 44.05% of respondents indicated "agree" or "strongly agree," showing nearly half hold positive attitudes toward Taiwan's society's crisis response capabilities. However, the proportion holding negative attitudes is also quite high at 45.02% (disagree 29.79%, strongly disagree 15.23%), slightly exceeding those who agree, indicating insufficient confidence in society's crisis response.

Short Videos and Influencer Impact

• Short videos and influencer content have become an undeniable part of contemporary life in Taiwan. From entertainment and leisure to knowledge acquisition, from daily sharing to political discussion, these emerging digital media forms are redefining the rules of information dissemination. However, as usage becomes widespread, related discussions and concerns are gradually emerging.

Short Video Usage Habits

• This survey first asked about the frequency of watching short videos from platforms such as TikTok, IG Reels, Facebook Reels, and YouTube Shorts over the past three months. Nearly 80% of Taiwan's population (78.46%) had experience watching short videos in the past three months, and nearly 60% (58.04%) watch short video content from various online platforms "every day." This result also shows that short videos have become an important component of daily information consumption for most people in Taiwan.

Expected Outcomes of Short Video Use: Forgetting Worries

• 50.72% of short video users tend to agree that scrolling short videos helps them forget their worries; conversely, 45.85% of users tend to disagree with this statement.

Expected Outcomes of Short Video Use: Encountering New Things

• Regarding information exploration, users' evaluation of short videos is relatively positive. Nearly 70% (69.17%) of short video users tend to agree that scrolling short videos allows them to encounter novel things they never thought of; conversely, those who disagree comprise only 27.64%, including 21.66% "disagree" and 5.98% "strongly disagree."

Guilty Pleasure from Short Videos

• This survey asked about emotional responses after using short videos. Regarding the degree of regret after spending time scrolling short videos, 47.07% of short video users answered "never," the highest proportion. Those with notably stronger regret—"frequently" and "always"—comprise only 5.77% and 2.23% respectively. Overall, about 60% indicate they rarely or never feel regret after use, showing most users' evaluation of post-use experience tends toward positive, with less obvious guilt or regret.

• In contrast to feelings of regret, the distribution of happiness after using short videos shows a different pattern. Regarding frequency of feeling happy after spending time scrolling short videos, the middle option "sometimes" has the highest proportion at 44.02%; those on the lower happiness frequency side comprise 29.59% ("never" 10.40% and "rarely" 19.19%) and those on the higher happiness frequency side comprise 25.58% ("frequently" 18.37% and "always" 7.21%), with close proportions. This result shows that public evaluation of happiness feelings after short video use is more dispersed, unlike regret feelings which concentrate at the low-frequency end, reflecting that positive emotional experiences from short videos vary by individual.

Overconsumption of Short Videos

• Beyond emotional experiences, whether overconsumption of short videos affects time management is also an important discussion dimension. When asked whether time spent scrolling short videos often exceeds original expectations, 56.09% of short video users express agreement; those disagreeing comprise 42.40%. This result shows over half of users admit difficulty controlling usage time according to expectations, reflecting that short video characteristic mechanisms (such as autoplay, algorithmic recommendations) may pose challenges to users' time management.

• Next, examining whether short video usage behavior has formed habits among Taiwan's population. 56.38% indicate they often start scrolling short videos unconsciously ("agree" 39.59%, "strongly agree" 16.79%), notably higher than the 42.45% who tend to disagree. This reflects that for a considerable proportion of the population, short video use has formed a certain degree of habitual response, automatically opening without deliberate thought, and this automatic behavior may further exacerbate the aforementioned time control difficulties.

• This survey also asked whether short video use causes substantial impact on daily life. Results show 62.27% believe they won't delay other matters due to watching short videos. However, over 30% (36.22%) admit they indeed experience problems of delaying other matters due to short videos, reflecting that some users' short video consumption has produced perceptible negative impacts on daily routines.

Frequency of Receiving Current Events Through Influencers

• Today there are many "influencers," referring to people who have fame online, regularly publish content online, and have a certain number of followers or fans. This survey asked about "having watched/listened to content published by influencers online in the past month," with over half (54.71%) having watched or listened to influencer-published content in the past month, while 30.58% had not. This shows influencer-published content has become one of the important channels for many people to receive information.

• To understand public behavioral patterns of receiving information through influencers, first regarding "knowledge"-related content conveyed by influencers, asking those who watch influencer content in the past three months: "How often do you watch or listen to content published by influencers related to knowledge such as travel and food, beauty and fashion, sports and healthcare, or popular science?" Results show "frequently" has the highest proportion of viewers/listeners at 32.73%; followed by "sometimes" at 29.70%.

• Beyond knowledge-type content, this survey asked about frequency of watching "political current events" content published by influencers. Results show "rarely" has the highest proportion of viewers/listeners at 40.74%.
  This frequency distribution shows obvious differences from the aforementioned frequency distribution for watching knowledge-related content, indicating that among those who watch influencer content, most still primarily receive knowledge-type content such as travel and food, beauty and fashion, sports and healthcare, or popular science, while watching political current events content is relatively less frequent.

Perception of Knowledge and Information Simplification

• To explore public perception of knowledge/information simplification, this survey asked: "Do you agree that influencers explain knowledge-related issues in simple and understandable ways?" Results show that regarding "knowledge"-related issues such as travel and food, beauty and fashion, sports and healthcare, or popular science, 84.35% of influencer content viewers tend to agree that influencers explain knowledge-related issues in simple and understandable ways; conversely, those tending to disagree comprise only 13.25%, showing an obvious gap with those agreeing.

• This survey similarly asked regarding "political current events" issues whether people believe influencers explain in simple and understandable ways. Survey results show slightly different agreement distribution compared to knowledge issues. Regarding "political current events" issues, 68.51% of influencer content viewers tend to agree that influencers explain political current events-related issues in simple and understandable ways; those tending to disagree comprise 28.23%. Compared to survey results on "knowledge" issue simplification perception, this shows a larger proportion did not perceive information simplification occurring in influencers' political current events information dissemination process.

Fraud Risks and Parenting Challenges

Public Encounters with Fraud

• Fraud continues to threaten people's financial security and quality of life, with various fraud tactics constantly evolving alongside digital technology development. To understand the current situation of public encounters with fraud, this survey asked: "In the past three months, have you encountered fraud messages? (such as fraud SMS, phone calls, advertisements, or malicious URLs, etc.)" Results show 60.21% of Taiwan's population encountered fraud messages through various channels in the past three months, with 2.94% actually suffering losses from being deceived.

Current State of Digital Parenting

• With the proliferation of digital devices, the age at which children begin using 3C products has attracted attention, and "using 3C devices to pacify children" has become a common yet controversial practice in modern parenting. This section examines Taiwan's digital parenting situation through two dimensions: first investigating the age distribution when children under 12 first use 3C devices; second exploring the frequency with which parents use 3C devices to pacify children in public places or during busy moments, comparing actual behaviors of those with parenting experience against general public attitudinal perceptions.
• This section first asked: "At what age did children under 12 in your household begin watching or using 3C devices (including phones, tablets, computers)?" to understand the overview of children's 3C device usage. Results show that among parents with children under 12, 74.70% allow their children to use 3C devices, with the preschool stage (0-6 years) showing the highest proportion at age 3, at 8.72%. Meanwhile, 25.30% of surveyed parents indicated children under 12 in their household do not use 3C devices.

• This survey further asked parents with children under 12 who have 3C usage experience: "When you are in public places or busy, and want your child to quiet down, how often do you let your child use 3C devices?" to understand how technology products like 3C devices are used as tools to pacify children in modern parents' child-rearing process. Survey results show 73% of parents with children under 12 who have 3C usage experience allow their children to use 3C devices to pacify them when necessary.

Expert Interviews: Digital Empowerment and Reflections in the Age of Artificial Intelligence

• In the wave of global digital transformation, the integration and application of technology has become a key force driving organizational progress, social advancement, and economic innovation. To understand the impact created by digital technology at the individual, organizational, and social levels, this research invited five scholar-experts to participate in discussions. The invited experts include academic researchers (Professor Fu-Ren Lin, Professor and Director of the Institute of Service Science at National Tsing Hua University; Professor Wen-Wei Shiu, Professor and Dean of the College of Liberal Arts at National Taiwan Normal University), industry practitioners (Pei-Jun Ho, Founder of Townway Cultural and Creative Corporation; Eric Chang, COO of Taiwan AI Labs), as well as technology practitioners spanning both industry and academia (Ren-Hao Pan, Founder of WaCare Telehealth and Adjunct Assistant Professor at National Yang Ming Chiao Tung University). They specialize in different fields of digital technology development and share professional insights on digital empowerment from diverse perspectives including their own experiences and research.
• The interviewed experts unanimously pointed out that the rise of AI has profoundly influenced the landscape of digital empowerment. AI's rapid development relies not only on the accumulation of algorithms and data training, but also on the maturation of cloud computing, Internet of Things, big data, and high-speed communications. Therefore, although this report centers on AI, the interview content actually demonstrates that AI's progress is not a miracle of a single technology, but the result of multiple technologies resonating together. Its popularization symbolizes not only technological advancement, but also the expansion of human intelligence boundaries and the redefinition of capabilities.
• Overall, the interviewed experts generally believe that the rise and rapid development of AI has deepened the practice of digital empowerment in Taiwan society. For example: in the education field, students use it as a learning partner while teachers leverage it to advance teaching and research efficiency; in daily work, AI helps spark inspiration and generate content; in business decision-making, AI extracts key content from complex data to assist humans in making more precise judgments. These changes enable individuals to better master information, organizations to become more innovative, and society to move toward a more equitable and inclusive direction. In fields such as education, business, healthcare, industrial upgrading, or public governance, AI has redefined the relationship between humans and technology and accelerated the popularization of intelligent technology.
• However, such development remains full of variables. The interviewed experts, from individual, organizational, and social perspectives, offered their insights on AI development and raised many profound reflections.

Individual Level: From Users to Collaborators

• AI's rapid development has brought technology closer to daily life. Especially with the popularization of generative tools, technology has become democratized and conversational. Through natural language interaction, multimodal learning, and prompt engineering, individuals are no longer merely users pressing buttons or operating interfaces, but collaborators working alongside AI: students use it to explore knowledge and formulate questions; creators leverage it to generate inspiration and produce drafts; workers use it to organize information and extend thinking. Technology has become an entity for collaborative thinking.
• However, interviewees also noted that this convenience comes at a cost. Does increased reliance on AI potentially weaken human creativity and judgment? Does greater dependence on digital technology more easily lead to social alienation and anomie? This so-called "AI Paradox" centers on the core issue that while AI extends human capabilities, it may simultaneously redistribute agency between humans and technology, potentially affecting the autonomy and depth of human thinking in the long term.

Organizational Level: From Process Automation to Innovation Ecosystems

• AAI has become a key driver of digital transformation for enterprises and organizations. It enhances the immediacy and precision of decision-making, promotes cross-departmental collaboration and knowledge circulation, and shifts organizational culture toward innovation models centered on data and intelligence. On the industry side, the software-hardware integration of Edge AI, cross-organizational federated learning, and predictive analytics for business decisions are accelerating. In education, AI-assisted teaching makes individualized learning more feasible. In society, AI also supports rural communities in establishing their own local resource platforms. AI is becoming an important engine driving value innovation.
• However, organizations simultaneously face another challenge: AI's black-box computing limits our understanding and oversight of its results. When models cannot be fully explained, how should practical applications and responsibilities be defined? How can transparency and accountability mechanisms become future governance models? AI provides efficiency but also requires organizations to rethink governance structures and ethical boundaries.

Social Level: From Technical Enhancement to Intelligent Shared Prosperity

• At the public service and social development level, AI demonstrates greater and broader possibilities. Combined with IoT and data analytics, healthcare becomes more preventive and immediate; generative and multimodal technologies assist language revitalization and cultural preservation; accessible communication tools enable people with disabilities to express themselves and participate more freely. From the perspective of societal impact, it becomes an emerging carrier of public value.
• However, this development direction remains full of uncertainty. Interviewees raised many thought-provoking questions: Does AI popularization still only benefit the minority with capabilities or resources? If education overly emphasizes AI skills, does it reduce students' willingness and possibility to explore other talents? If corporate resources are heavily invested in AI, does it squeeze other industries' survival space? Could the centralization of computing power and datasets create a new "digital feudalism" phenomenon?
• While AI opens infinite possibilities, it also reflects deep challenges. In these interviews, expert interviewees discussed not only technological development but also contemplated what kind of AI society we want to become. What kind of existence do we want AI to be in order to form a prosperous and harmonious society? They offered valuable recommendations.

Bridging the Digital Divide: Making Technology a Force for Everyone

• If AI belongs only to a few, technological progress will widen the existing digital divide and become a new form of inequality. School and social education, along with corporate training, must keep pace with technological advancement, enabling everyone to have the capability or imagination to participate in this transformation. Digital learning cannot be merely knowledge input on screens, but must grow through human-to-human interaction in real-world contexts. Only when AI application aligns with organizational culture can organizational digital transformation truly become an extension of humanity rather than new pressure, with organizations pursuing process changes where AI and employees thrive together. Building AI as "borderless digital penetration" enables urban-rural mobility and intergenerational shared prosperity, making technology not a divide of knowledge or geography, but a connection.

Confronting the AI Paradox: Reawakening Human Thinking and Critical Capacity

• Greater reliance on AI deepens our doubts and concerns. Experts remind us that only by re-cultivating individual human agency can we maintain authentic happiness in the digital world. When people reconnect with real communities and self-worth, digital isolation and addiction can be reduced. The more courageously people engage with strangers and the more capable they are of authentic self-expression, the greater the opportunity to dissolve invisible alienation. Meanwhile, they also call for recognition that while AI technological development is important, it should not be the only path to success. Education should provide more diverse pathways and choices, allowing young people to still see other possibilities when facing setbacks.

Responsible AI: Allowing AI to Mature Through Transparency and Accountability

• Data privacy and ethical governance are core issues in AI applications. AI model training relies on vast amounts of personal and organizational data. If data acquisition, usage, and storage processes lack transparency, trust crises will result. To avoid algorithmic bias and ensure decision-making fairness, experts believe traceable regulatory and review mechanisms must be established to positively enhance user acceptance and trust in AI. Developing "explainable AI" and "responsible AI" is not only a global consensus but also an imperative Taiwan cannot ignore when facing an intelligent society.