Re-design ChatGPT

Walushka Faleiro
Walushka Faleiro
    • Get in touch
    Download color palette

    Grounding in communication

    To help grounding in communication quite some features were added. Role selection helps users and ChatGPT agree on the context of the conversation they have and the users know what to expect alongside ChatGPT to know what to provide. Personalization forms created facilitate mutual understanding by all the data provided by the user with consent. Multilingual support helps ChatGPT interpret the user's voice accurately avoiding miscommunication and by comprehensive understanding of users prompts, since people from diverse cultures participate in one-to-one or one-to-one communications. Syntax defined as ChatGPT’s language, where the user can add new commands to create a common grounded language with consistent outcomes. Functionality, where ChatGPT attends calls and answers texts, helps reduce errors in repetitive or routine communication. For eg, the user wants to message 10 people about a new launch of the product. Turn-taking during group conversation was also added.

    Hindrance in Grounding

    One main issue can be cognitive overload, where the user has to assign roles and learn the syntax and new users can struggle initially. These hindrances were resolved in the following ways. Firstly, the user is given the option of creating their own commands by voice recording. During conversations, ChatGPT suggests syntaxes to reduce effort in typing prompts and ease of use of functionalities. With respect to role selection, if the user doesn't select any roles, the automatic general mode is set as well as ChatGPT's suggests the role that should be assigned when conversing, which can be changed in the chat directly. ChatGPT might not be able to Automate calls and texts exactly as wanted, hence, the instructions field was added allowing the addition of key details or exact messages to be sent. Another innovative idea was letting ChatGPT inform users when it's unsure of contexts even a bit. All these strategies might still not support grounding fully, but will reduce misunderstandings. Brennan, S.E. (1998) provided insights on grounding during communication with computers.

    Miscommunication

    ChatGPT uses its ability to display 3-D projection, uses a mindmap to illustrate the discussion had by people, dates the discussions took place, and the proximity of bubbles to show the participants were aligned and with whom. A small summary card containing the possible issue found is also projected . The participants can further click on the bubbles and see the tree structure of the texts.

    Group Communication

    During group conversations, ChatGPT analyses the texts to know when to reply to avoid Interrupting. Besides the group name (study group), the general option selected means ChatGPT is set to the general mode which is automatic.

    ChatGPT’s can be assigned specific roles, to perform its duty to its best ability. In this conversation as suggested by ChatGPT, it is assigned the teacher's role, further made known to all group members.

    For the rationale of turn-taking during conversing, the Chat is highlighted with the application colour (purple) by ChatGPT to inform it’s going to text as well as in the top centre, who is typing is displayed. This facilitates to know the duration someone is still texting.

    Detailed description of new elements added

    Role Selection: This functionality allows the user to select the role the chatGPT will play during the conversation in the chat. Considering that Chatgpt can already support writing, medical sector information, Education sector information, sales sector, as a virtual assistant, history although it's not that accurate yet. Personalization and user experience could be highly enhanced. For eg is ChatGPT was a teacher in a chat, and the user sent an essay, Chatgpt could be given the ability to behave like a teacher, suggesting improvement not just in the content but also improving ways of getting better at writing essays, giving some work as a practice and provide a schedule to learn a particular topic to get better at essay writing.

    Add your magic: Since ChatGPT doesn't have the ability to accurately process natural language processing. Develop an emotional connection with users to build trust, this way the process can be facilitated faster. This also includes the consent provided for tracking and choosing whose calls and chats ChatGPT is allowed to answer.

    Activity Tracker: Metrics tracked– Happy, Sad, Annoyed. Work done, and work pending (if access is given) could be shown with a sliding bar, deadlines missed, time off taken, skill improvement based on the way data is given to ChatGPT. Based on the responses of the user with ChatGPT, ChatGPT can conduct a sentimental analysis. The sole purpose of this functionality is for the user reflection for personal growth.

    Co-creation: This adds a little zest to the working of ChatGPT. Where you can get the best of ChatGPT by speaking the language ChatGPT knows best and receiving results faster. Template prompts, an idea inspired by metacognition. Where ChatGPT asks the user further questions before answering for accuracy. Users can also use examples of templates provided to get the right result. In this, the user should answer question like  what, why, when, and how preventing excessive dependence. To support personnel growth, the user should be given an opportunity to give the initial ideas or solution in the prompt.

    Speak for you: If you want to attend the meeting but you’re sick you can allow ChatGPT access to teams. Following this, it will provide you with a detailed summary and during meetings it can send texts accurately to let others know you are present.

    Multi-Language support: Chatgpt can converse in any language the user is comfortable in and can even learn the language pronunciation, and the communication style as the user converses, giving them the vibe of a friend or native person.

    Confidential mode: When this mode is switched on, Chatgpt assures the user that their data is stored in a particular part that is not accessible to anyone or shared by anyone. It assures users that no data is passed on except what is widely available on the internet. 

    Adding Chatgpt to Applications: An option to allow Chatgpt to various other Apps used to work like Gmail, Google Docs, and Canva to facilitate co-creation and  support during work. Although this raises concerns of privacy, exact document access, and particular function access like Grammarly could also be given. For example, when user gives access to their only work email in Gmail to tackle privacy issues, ChatGPT could filter and organize the emails giving a clear summary and notifying them before certain events to not miss out on crucial events at work or to get things done on time. 

    Prototype Link

    https://drive.google.com/drive/folders/1sGepPBXr_GHzL66CH8Q2NZv5PID6mIGB?usp=sharing

    That’s a wrap on my ChatGPT redesign! Hope you enjoyed the journey—whether you loved it, laughed at it, or got inspired. Drop your thoughts below, or just send a virtual high-five! 🙌😆

    References

    1. Ashktorab, Z., Jain, M., Liao, Q. V. and Weisz, J. D. (2019) 'Resilient chatbots: Repair strategy preferences for conversational breakdowns', in Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems. New York, NY: Association for Computing Machinery, pp. 1–12. Available at: https://doi.org/10.1145/3290605.3300484 (Accessed: 6 January 2025).

    2. Amershi, S., Weld, D., Vorvoreanu, M., Fourney, A., Nushi, B., Collisson, P., Suh, J., Iqbal, S., Bennett, P.N., Inkpen, K., Teevan, J., Kikin-Gil, R. and Horvitz, E. (2019) ‘Guidelines for human-AI interaction’, Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, Glasgow, Scotland, UK, pp. 1–13. Available at: https://doi.org/10.1145/3290605.3300233(Accessed: 20 December 2024).

    3. Brennan, S.E. (1998) 'The grounding problem in conversations with and through computers', in Social and Cognitive Approaches to Interpersonal Communication. 1st edition. Psychology Press, pp. 25. Available at: https://doi.org/10.4324/9781315805917 (Accessed: 20 December 2024).

    4. Design, L. (2022) Purple Blue Galaxy Color Palette: Lilybug design, Pinterest. Available at: https://uk.pinterest.com/pin/325948091792484249/ (Accessed: 06 January 2025).

    5. Fatima, I. (2023) Designing with AI: A user study to explore the future role of AI as a collaborative tool in graphics design (Dissertation). Södertörns University. Available at: https://urn.kb.se/resolve?urn=urn:nbn:se:sh:diva-52673 (Accessed: 6 January 2025).

    6. Halliwell, J. (2024) '‘Are you ready for a little chat?’: Creating and developing WhatsApp group chats as a digital research method', Area. Available at: https://doi.org/10.1111/area.12986 (Accessed: 6 January 2025).

    7. Lazuardi, S. (2020) Handz 3D illustration | figma, Figma.com. Available at: https://www.figma.com/community/plugin/856130306757736209/handz-3d-illustration  (Accessed: 24 January 2025).

    8. Mateusz Brodowicz (2024) The Impact of Typographical Errors on Effective Communication. Business, Education. Published on 16 July. Available at: https://www.example.com (Accessed: 23 January 2025).

    9. Nouwens, M., Griggio, C.F. and Mackay, W.E. (2017) ‘"WhatsApp is for family; Messenger is for friends": Communication places in app ecosystems’, Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, Denver, Colorado, USA, pp. 727–735. Available at: https://doi.org/10.1145/3025453.3025484 (Accessed: 20 December 2024).

    1. Pan, B. et al. (2024) AgentCoord: Visually exploring coordination strategy for LLM-based multi-agent collaboration, arXiv.org. Available at: https://arxiv.org/abs/2404.11943 (Accessed: 20 January 2025). 

    2. Premium vector | audio recording app mobile UI recorder design (no date) Freepik.com. Available at: https://www.freepik.com/premium-vector/audio-recording-app-mobile-ui-recorder-design_33909069.htm (Accessed: 06 January 2025).

    3. Santhosh, R., Abinaya, M., Anusuya, V. and Gowthami, D. (2023) 'ChatGPT: Opportunities, Features and Future Prospects', 2023 7th International Conference on Trends in Electronics and Informatics (ICOEI), Tirunelveli, India, pp. 1614–1622. Available at:https://doi.org/10.1109/ICOEI56765.2023.10125747(Accessed: 2 January 2025).

    4. Seufert, M., Hoßfeld, T., Schwind, A., Burger, V. and Tran-Gia, P. (2016) ‘Group-based communication in WhatsApp’, 2016 IFIP Networking Conference (IFIP Networking) and Workshops, pp. 536–541. Available at: https://doi.org/10.1109/IFIPNetworking.2016.7497256 (Accessed: 20 December 2024).

    5. Shahriar, S. and Hayawi, K. (2023) “Let’s Have a Chat! A Conversation with ChatGPT: Technology, Applications, and Limitations”, Artificial Intelligence and Applications, 2(1), pp. 11–20. Available at:https://doi.org/10.47852/bonviewAIA3202939(Accessed: 2 January 2025).

    6. Tankelevitch, L., Kewenig, V., Simkute, A., Scott, A. E., Sarkar, A., Sellen, A. and Rintel, S. (2024) 'The Metacognitive Demands and Opportunities of Generative AI', Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (CHI ’24), Honolulu, HI, USA. Available at:https://doi.org/10.1145/3613904.3642902 (Accessed: 6 January 2025).

    7. Trushkin, V. (2019) Iconify | Figma, figma.com. Available at: https://www.figma.com/community/plugin/735098390272716381/iconify (Accessed: 24 January 2025).

    8. Wang, D., Maes, P., Ren, X., Shneiderman, B., Shi, Y., and Wang, Q. (2021) ‘Designing AI to Work WITH or FOR People?’, Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems. New York, NY, USA: Association for Computing Machinery. Available at: https://doi.org/10.1145/3411763.3450394(Accessed: 28 November 2024).

    9. Wang, D., Churchill, E., Maes, P., Fan, X., Shneiderman, B., Shi, Y., and Wang, Q. (2020) ‘From Human-Human Collaboration to Human-AI Collaboration: Designing AI Systems That Can Work Together with People’, Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems. New York, NY, USA: Association for Computing Machinery. Available at: https://doi.org/10.1145/3334480.3381069.(Accessed: 28 November 2024).

    1. Weisz, J.D., He, J., Muller, M., Hoefer, G., Miles, R. and Geyer, W. (2024) ‘Design principles for generative AI applications’, CHI ’24, May 11–16, Honolulu, HI, USA. ACM. Available at: https://doi.org/10.1145/3613904.3642466 (Accessed: 6 January 2025).

    2. Wei, Y., Lu, W., Cheng, Q., Jiang, T., and Liu, S. (2022) ‘How humans obtain information from AI: Categorizing user messages in human-AI collaborative conversations’, Information Processing & Management, 59(2), 102838. Available at: https://doi.org/10.1016/j.ipm.2021.102838(Accessed: 28 November 2024).

    Walushka Faleiro
    Walushka Faleiro
    Get in touch

    More by Walushka Faleiro

    View profile
    • Like