3 Papers I Wish I'd Written

Thank you to Professor Kai Lukoff for providing template "3 research papers that I wish I had written," which have helped me delve into research reading and discover my main interest in HCI.

This article lists my top 3 recently read research papers on HCI. Essentially, they relate to innovative HCI applications using Generative AI.

Paper1: Closer Worlds: Using Generative AI to Facilitate Intimate

Papaer Title: Closer Worlds: Using Generative AI to Facilitate Intimate

Link: https://dl.acm.org/doi/10.1145/3544549.3585651

Key research question(s):

1. How do digital communication tools affect our ability to experience deep emotional intimacy, and can these tools be improved to better foster a sense of connection?

2. Can games and generative AI art be used to counteract the trend of digital communication tools limiting emotional intimacy?

3. What design principles, inspired by facilitation methods, can be effective in fostering emotionally intimate conversations in a ML-assisted 2-person game?

4. How effective is the "Closer Worlds" game in eliciting self-disclosure and fostering emotional intimacy compared to social games without generative AI?

5. Do the affordances provided by visualizing shared values through a co-creative game lead to comfortable and meaningful conversations?

Key Methods:

1. Design and Development of "Closer Worlds":

   • The researchers have created a machine learning-assisted game for two players intended to encourage emotionally intimate conversations through co-creative world-building.

2. Exploration of Design Principles:

   • The paper discusses design principles derived from facilitation methods that may help in encouraging emotional closeness within the game environment.

3. Pilot Study:

   • A pilot study is conducted to assess the effectiveness of the game. This involve interviews, observations, surveys, or behavioral analysis, to measure self-disclosure and emotional intimacy.

4. Comparative Analysis:

   • A comparison is made between Closer Worlds and a social game without generative AI to see how each affects the level of self-disclosure among participants.

5. Assessment of Enjoyment and Novelty:

   • The researchers also assess the participants' enjoyment and the novelty of the experience provided by the game, which can influence the game's effectiveness in fostering intimacy.

6. Discussion on Future Applications:

   • Finally, the paper discusses how co-creative games might leverage generative techniques in the future to create pro-social environments.

Paper2: WorldSmith: Iterative and Expressive Prompting for World Building with a Generative AI

Papaer Title: WorldSmith: Iterative and Expressive Prompting for World Building with a Generative AI

Link: https://dl.acm.org/doi/10.1145/3586183.3606772

Key research question(s):

1. How can multi-modal image generation systems be used to assist in the process of fictional worldbuilding?

2. What are the limitations of current "click-once" prompting UI paradigms in generative AI, and how can they be improved to benefit the creative process?

3. Can a system like WorldSmith enable novice worldbuilders to visualize their concepts effectively and with greater ease?

4. How do iterative visualization and modification techniques (text input, sketching, region-based filling) impact the creative process in fictional worldbuilding?

Key Methods:

1. Development of the WorldSmith Tool:

   • The paper discusses the creation of a tool that combines text input, sketching, and region-based filling for users to visualize and edit their fictional worlds.

2. Formative Study:

   • A preliminary study involving 4 participants was conducted to gather initial feedback on the tool's functionality and usability. This involve qualitative methods such as interviews and observations.

3. First-Use Study:

   • A larger study with 13 participants was then performed to see how new users interact with the WorldSmith tool. This involve a mix of qualitative and quantitative methods, including task performance measurements, questionnaires, and user feedback sessions.

4. Expressive Interactions Analysis:

   • The study includes an evaluation of how the tool allows for expressive interactions between the users and the prompt-based models.

5. Comparative Analysis with Existing UI Paradigms

6. User Empowerment Assessment

Paper3: DeepScope: HCI Platform for Generative Cityscape Visualization

Papaer Title: DeepScope: HCI Platform for Generative Cityscape Visualization

Link: https://dl.acm.org/doi/10.1145/3334480.3382809

Key research question(s):

1. How can the process of creating high-quality streetscape visualizations be improved to support urban design and planning?

2. Can a Generative Neural Network (DCGAN) combined with a Tangible User Interface (TUI) streamline the visualization process for urban design, especially in real-time, multiparty design sessions?

3. What are the design, development, and deployment considerations for a platform like DeepScope?

4. How does DeepScope potentially impact the urban design process, and what are its practical implications?

Key Methods:

1. Design and Development of DeepScope:

   • The paper describes the conceptualization and creation of the DeepScope platform, detailing the integration of a DCGAN with a TUI.

2. Technical Implementation:

   • The implementation phase involves the actual coding, algorithm development, and hardware setup necessary to create a functioning prototype.

3. Real-Time Urban Planning Simulation:

   • DeepScope is utilized in a simulated environment to assess its performance in real-time urban planning scenarios. This involve the use of real-time feedback systems.

4. Multi-Participant Testing:

   • The system is tested in an environment with multiple participants to simulate a real-world urban design session.

5. Feedback and Iteration

6. Case Studies or Deployment Narratives

7. Discussion and Analysis