ChatGPT, developed by OpenAI in 2022 and with 180.5 million users worldwide, has transformed how we interact with technology to facilitate our lives. However, as it stands, it offers few features compared to other free AI chatbots. This project focuses on introducing features to enhance ChatGPT’s user experience in the growing AI sector.
Users struggle to craft prompts on ChatGPT and spend time editing content to meet their needs, leading to frustration and a feeling of misuse. This not only impacts user satisfaction, but impacts ChatGPT as competitors with a more user-friendly experience emerge.
The following proposed ChatGPT features to provide a more personalized ChatGPT experience. Guidance in writing effective prompts and streamlined content organizing aims to empower users and solidify ChatGPT’s position as a reliable and valuable tool in the AI market.
ChatGPT’s launch sparked a wave of competitors. Analysis of Copilot, Gemini, Claude, and Perplexity revealed that compared to these competitors, ChatGPT’s free version provides limited features. Based on this, I believe that to retain users ChatGPT’s free version has to offer features that add more functionality and usability to the tool.
To discover what features would benefit users, it was important to understand their relationship with ChatGPT. User interviews and contextual observation of 7 daily users interacting with the tool brought to light user frustrations with prompt engineering, tone of responses, and ChatGPT’s user interface. It was surprising to learn that users blamed themselves for subpar responses, feeling they lacked the expertise to use ChatGPT effectively.
How might we guide and personalize the ChatGPT experience to empower users and reinforce its role as a reliable optimization tool?
Goals were set to help keep brainstorming focused on features that provide the most value to ChatGPT and users. To retain users, features had to focus on keeping ChatGPT competitive with other free AI Chatbots and providing an empowering, user-friendly experience.
Must-haves, Should-haves, Could-haves, and Won't-haves. This approach focused on delivering maximum value to ChatGPT and its users, emphasizing features that enhance user experience and maintain competitiveness with other free AI chatbots while reinforcing ChatGPT's reliability as an optimization tool.tures that foster connection, encourage meaningful engagement, and stand apart from existing tools.
The main actions introduced with the new features are selecting suggested and predictive prompts, creating a group, adding a chat to a group, and customizing settings. Working on user flows alongside low-fidelity wireframes of the proposed features allowed for easier visualization of each step, identifying key screens and potential pain points.
Mid-fidelity designs were created to arrange the feature designs and user flows. These designs were evaluated for clarity and usability of proposed features. Feedback from this evaluation was incorporated into high-fidelity designs, reflecting a call for increased clarity and more streamlined user flows.
High-fidelity designs were tested with 10 participants through moderated usability tests. The results showed an increase in usability and a positive perception of the features but there was room for improvement.
User feedback points to a lack of intuitiveness for accessing the Set Context modals through the existing “Customization Settings” icon in the main text input area. However, after careful consideration of the novelty of both ChatGPT and the proposed features, user adoption is expected. Placing the icon within the main text area makes the icon a prominent part of the design, encouraging user interaction through familiarity. Over time, it is expected that users will learn the icon’s function.
This project taught me about the importance of assessing feedback to determine what is actionable. When assessing feedback regarding unfamiliar icons and feature placement, I had to weigh it against ChatGPT’s brand and the novelty of the tool. Should these features exist in ChatGPT, my hypothesis would be that they would over time become familiar and valuable to users. As with any hypothesis, further testing would be needed to validate it.It also taught me to leverage existing design elements. The Create Group and Set Context modals were the most difficult components to design. I gained some clarity when I realized I could expand on existing components (Customization Settings and its icon) instead of designing new ones. This approach to designing the modals ensured that over time users become more familiar with them and helps keep the design system clean and scalable.
