"AI Afterlives" as Digital Legacy: Perceptions, Expectations, and Concerns

Research Background and Issues

• Identified Problems or Challenges:

  1. With the rapid development of generative AI technologies, people have begun using digital information to create an "AI Afterlife" as a form of digital legacy. This technology is not merely a static digital record but involves dynamic, interactive, and generative content, offering the possibility for individuals to communicate with the world after death.
  2. However, there is currently limited human-centered research, particularly from the perspective of individuals being simulated by AI agents, to systematically analyze perceptions, expectations, and potential issues surrounding this form of digital legacy.
  3. Major challenges include unresolved ethical and social norms of digital legacies, identity consistency issues triggered by "AI Afterlife" simulations, and potential disturbances in interactions with individuals and families.

• Significance of the Research:

  1. Traditional digital legacies (e.g., social media accounts, passwords, files) are typically static, whereas the "AI Afterlife" introduces dynamic and interactive features, providing mourners with deeper emotional connections or long-term value.
  2. The development of this technology involves complex challenges related to personal privacy, ethical norms, and societal acceptance.
  3. Understanding users' needs, expectations, and concerns regarding the "AI Afterlife" is crucial to avoiding potential legal, social, and ethical issues.

• Research Motivation and Related Work:

  1. Research on traditional digital legacies has primarily focused on the storage and inheritance of static content, while studies on forms that continuously generate new information and interact with the environment are still in their infancy.
  2. Previous studies have shown that generative AI can help individuals establish a "continuing bond" during the grief of losing loved ones, but there is a lack of exploration of the "AI Afterlife" from the perspective of the represented individual.

Proposed Solution

• Methods and Innovations:

  1. The authors conducted semi-structured interviews with 18 participants from diverse demographic backgrounds to deeply explore their perceptions, expectations, and concerns regarding the "AI Afterlife."
  2. The innovation lies in extending the lifecycle of traditional digital legacies (encoding, access, processing) to the "AI Afterlife" and integrating the interactive characteristics of generative AI to inspire design.
  3. The authors proposed key design considerations for maintaining identity consistency and balancing between comfort and intrusion.

• Implementation Steps and Key Technologies:

  1. Data Collection: In-depth interviews were conducted to address lifecycle design issues (encoding, access, and processing) and interaction design for the "AI Afterlife."
  2. Data Analysis: Thematic analysis was used to extract core themes from participants' interview data, including factors influencing attitudes (personal, family, technological, and societal levels), comparisons with traditional legacies, interaction design, and technological limitations.
  3. The study extensively explored the need for AI agents to possess authenticity, evolutionary capabilities, and dynamic response mechanisms.

Research Outcomes

• Specific Findings:

  1. Identified key factors influencing participants' attitudes toward the "AI Afterlife," including personal views on life and death, family relationship needs, technological capabilities, and societal perceptions.
  2. Explored three major distinctions between the "AI Afterlife" and traditional digital legacies: data integration versus dispersion, content authenticity versus generativity, and interactivity versus static nature.
  3. Proposed a comprehensive lifecycle model and interaction design framework for the "AI Afterlife," including design improvement suggestions for agent identity consistency and interaction intrusiveness.
  4. Defined major concerns surrounding technological limitations, mental health risks, security, and economic issues.

• Advantages:

  1. The "AI Afterlife," through generative AI technologies, offers real-time interaction and information integration capabilities, fundamentally surpassing the static nature of traditional legacies.
  2. This research provides a human-centered methodology, establishing clear guidelines for users throughout the creation and interaction processes of digital legacies.

• Experimental or Evaluation Results:

  1. Participants exhibited dual attitudes toward the "AI Afterlife": some believed it could extend value and family connections, while others were concerned about technological inauthenticity and potential ethical issues.
  2. The study revealed that most participants desired to maintain core identity traits (appearance, knowledge, thoughts) in the "AI Afterlife," along with controllable evolutionary capabilities to adapt to dynamic environments.

• Limitations and Future Directions:

  1. The study's participants were primarily from a Chinese cultural background, where views on life, death, and family honor significantly influenced the findings. Future research should validate these results in broader cultural contexts (e.g., Western individualistic societies).
  2. Technological and interaction designs require further optimization to effectively balance identity consistency management and family support functions.
  3. Interdisciplinary research is needed to address ethical, legal, regulatory, and societal acceptance frameworks.

Through this study, the authors not only deepened the understanding of the potential of the "AI Afterlife" in the digital legacy domain but also proposed important design theories and practical insights for optimizing the lifecycle and interaction experience of digital legacies.