There are increasing concerns regarding an insidious consequence of social media: the erosion of user autonomy. The research aims to explore unreflective habit formation – an expression of loss of user autonomy - in the context of TikTok usage in Vietnam and to apply STAMP model to conceptualize the issue and the STPA method to provide actionable recommendations to the Vietnam Government. Being explorative, the research adopts desk research and qualitative methods. By integrating different concepts of autonomy in philosophy and HCI, the research identifies five key dimensions of autonomy, namely decision, execution, self-congruence, independence and system hostility. As social media exploits neurological reward cycles to reinforce user habits, user autonomy is infringed under the decision, execution and system hostility dimensions. Semi-structured interviews were conducted with 12 Vietnamese female university students in Hue during April 2025, revealing that their impulsive TikTok habitual use is unreflective. There are three main losses from this impulsive habit: time, focus capacity and identity rigidification. The responsibility for regulating social media platforms lies with the Government. They should adopt a combined strategy that integrates strict regulation of social media with comprehensive public education. These measures must be informed by up-to-date scientific research on user behavior and platform design. Regulations should not target TikTok in isolation, as the platform’s features are widely replicated across the industry. While applying STAMP model and STPA method to social media reveals both challenges and benefits, this approach is appropriate as it offers actionable insights that support targeted and comprehensive policy strategies.

The prevalence of mental health issues among adolescents and young adults is increasing, with social media usage identified as a potential contributing factor. As technology becomes more integral to daily life, the design decisions of social media platforms emerge as significant determinants of users' mental well-being.
Despite awareness of social media's general adverse effects, there is a gap in understanding the specific impact of different design features. Academic literature identifies features like likes and direct messaging as having nuanced effects on mental well-being, but lacks a comprehensive overview for developers to reference. This thesis seeks to bridge this gap, focusing on design features and their psychological impacts to propose design interventions for enhancing mental well-being.
Employing Value Sensitive Design (VSD) as an analytical framework, this study is the first to apply VSD to social media in the context of mental well-being. It involves a tripartite methodology — conceptual, empirical, and technical investigations — to identify significant design features and values related to mental well-being on social media platforms. A novel social media platform, Dime, designed with mental well-being in mind, serves as a case study for evaluating these features and proposing design requirements.
The study identifies 8 key design features affecting mental well-being and outlines 13 values divided into higher and lower levels, facilitating focus on critical values for desired outcomes. It explores value interrelationships and conflicts, particularly noting indirect tensions between values like authenticity and connection, and between personalization, control, and autonomy.
Proposed design interventions aim to resolve identified value conflicts, enhancing authenticity to support connection and enhancing personalization and control to support autonomy.

This thesis explores the challenges faced by students with vision impairment in accessing visual educational content, with the goal of designing an inclusive tool to enhance their learning experience. Adopting a human-centered design approach backed with value-sensitive design, the research methodology combined qualitative and quantitative techniques including interviews, surveys, and design ethnography. These methods provided insights into the lived experiences of visually impaired students, revealing difficulties in comprehending complex visual data and feelings of exclusion from the academic community.

The design process was structured around four key phases – Exploration, Ideation, Iteration, and Implementation. The Exploration phase involved in-depth user research to identify needs and values. Ideation generated numerous concepts, with a focus on converting visual content into accessible multisensory formats. Iteration prototyped and tested these concepts to refine the designs based on user feedback. Finally, Implementation delivered the high-fidelity prototype – an AI-powered platform called “InsightSpace” that integrates with the Brightspace LMS.

Key features of InsightSpace include the “InsightOut” tool which leverages AI to interpret and narrate visualizations, and the Envision Ally community portal that facilitates collaborative learning. The concept aligns with design values like accessibility, inclusiveness, personalization, and community identified through the research. Evaluation metrics focus on usability, user satisfaction, and platform effectiveness in conveying visual information.

The thesis concludes that InsightSpace demonstrates the potential of inclusive design and assistive technology in transforming educational experiences. It underscores the importance of multifaceted research and continuous user involvement in developing empathetic solutions.
The platform marks progress in the mission to create learning environments where students with vision impairment can thrive alongside their sighted peers.