Digital Empowerment Beyond Access: From Penetration to Inclusion
Taiwan's internet penetration reached 88.75% in 2025, entering a plateau phase. However, learning willingness among non-users plummeted from 8.23% in 2024 to 4.45% in 2025. Non-users face dual barriers: unfamiliarity with devices and low perceived need. When learning costs exceed expected benefits, abandoning adoption becomes rational. Notably, "age" has become the largest psychological barrier, with age-related resistance far outweighing any single application incentive. Meanwhile, 6.52% remain in a vulnerable position—needing assistance but lacking it—risking systemic exclusion in critical areas like healthcare and finance.
Policy thinking must shift from "increasing penetration" to "ensuring equitable participation." Digital-First should not mean Digital-Only. Physical service channels must remain as essential infrastructure for digital inclusion, especially for critical services like healthcare and taxation, with non-digital alternatives legally mandated. To address age barriers, adopt a dual approach: age-appropriate design to lower technical thresholds, and intergenerational learning programs to rebuild seniors' confidence. For vulnerable groups lacking family support, establish community digital service stations providing localized assistance and training to fill support gaps.
AI Empowerment Leap: From "Novelty" to "Utility," Yet "AI Literacy" Remains a Critical Gap
Generative AI has become the core engine driving digital capability upgrades. Usage of ChatGPT, Gemini and similar tools surged to 43.19% in the past three months, concentrated among young, highly educated, and northern urban populations. The critical finding lies in substantive transformation: once the usage threshold is crossed, self-efficacy rapidly develops. Among users, 70% feel capable of detecting AI, nearly 80% can evaluate its pros and cons, and 66.23% are confident in using AI to achieve goals—up 7.85 percentage points year-over-year. This confirms AI has evolved from novelty to a practical tool users confidently control for problem-solving.
However, as AI applications deepen, widening digital divides must be addressed. To prevent disadvantage among lower-educated and middle-aged to elderly groups, government should ensure "substantive equality." Short-term: offer digital transformation courses and workshops to help citizens become smart AI users. Long-term: integrate AI literacy into compulsory education, ensuring all demographic groups benefit from the AI wave.
Enhancing Public AI Risk Awareness: How Communication Framing Affects Vigilance Toward Misinformation and Privacy Breaches
In generative AI risk perception, concerns about "privacy breaches" significantly exceed those about "misinformation," reflecting that citizens prioritize tangible privacy and financial security over information authenticity. Interestingly, messaging tone significantly influences risk judgment. For "misinformation," definitive framing (emphasizing AI will produce errors) better raises risk awareness (20.14%), as information authenticity is hard to discern and requires assertive reminders. Conversely, for "privacy breaches," possibility framing (emphasizing AI may leak data) performs slightly better (39.47%). This shows that for personally relevant privacy risks, merely triggering suspicion suffices to generate vigilance, while misinformation requires clear authoritative warnings for effective alerts.
Challenges to Autonomous Empowerment: Short Videos Bring Positive Experiences But Erode Time Autonomy
Short videos have become habitual behavior, with 56.09% of users admitting usage often exceeds expectations, and a corresponding proportion unconsciously starting to scroll. Despite recognizing difficulty controlling time, 74.55% rarely or never feel regret. This "aware of loss of control but no regret" paradox reflects that platforms successfully extend user engagement through autoplay and algorithmic strategies while providing value through novel content discovery, yielding positive experiences for most users.
However, embedded risks cannot be ignored. Over one-third (36.22%) admit short videos have delayed other important tasks, potentially indicating problematic usage. For this group, digital literacy education needs targeted intervention strategies: establishing concrete time management mechanisms, cultivating self-awareness, or finding alternative emotional regulation methods to balance technological convenience with life order.
Crisis of False Empowerment: Nearly 60% Confident in Verifying Information, Yet 66% Never Verify
This year's survey reveals Taiwan's stark digital defense reality: 66% consider cognitive warfare threats serious, showing high societal awareness. However, this collective anxiety falls into a "high confidence, low action" paradox. While nearly 60% (57.11%) of internet users feel confident verifying news authenticity, this confidence fails to translate into actual verification habits—over 60% never or rarely proactively verify, relying instead on intuitive judgment.
This "confidence-action disconnect" stems from cognitive gaps regarding emerging threats. Most defense experience remains at the level of identifying crude traditional fake news, unable to detect that false information is now sophisticated packaged within neutral entertainment or educational content. Additionally, "resource scarcity perception" exacerbates defensive helplessness—citizens feel they lack effective, accessible resistance resources. Even when suspicious, they often abandon verification due to uncertainty about which tools to use or distrust of existing platforms. Bridging these cognitive and resource gaps will be the critical challenge for enhancing digital resilience.