The Project
Reimagine Labs is a startup developing a generative AI tool to assist the social and public sectors with research and business development. I created the wireframes for their MVP, which included templates and AI-assisted editing interactions.
The Challenge
The main challenge was designing a user experience that effectively gathered user information to inform the AI model, ensuring relevant and accurate content. This required an approachable interface that made users feel in control while the AI acted as an assistant.
The Solution
To address these challenges, I designed a platform that guides users through questions to collect personalized data, fine-tuning the AI model for relevant content. Key features include:
Personalized Data Collection: Users answer questions to gather necessary data.
Content Refinement: Users can edit AI-generated content with or without AI assistance.
Enhanced Editing Controls: Intuitive controls streamline the editing process.
Research and Insights
Conducted secondary research to understand existing food security initiatives and identify gaps.
Engaged in primary research through semi-structured interviews, surveys, and field observations to gain deep insights into the community's goals, motivations, and pain points.
Notion AI:
Stood out for its ability to highlight and edit specific text, offering AI-assisted options that enhance adaptability and user control. It helps users overcome prompting challenges and makes a wide range of options easily accessible.
Relume AI:
Allows users to edit specific text boxes, providing clear feedback on which sections are being edited and offering appropriate use case prompts. However, it lacks the ability for users to reject AI-generated content before replacing it or revert to previous edits.
How do we Design an effective AI Interaction?
Generative AI interaction is evolving from one-directional to a two-way conversation, thanks to advancements in Natural Language Processing. Users can now personalize AI instructions, provide real-time feedback, and specify desired outcomes, creating a more dynamic interaction. However, as user autonomy increases, so do their expectations for the AI to understand and anticipate their needs.
Adaptive systems theory offers principles to guide the design of effective AI products, ensuring they adapt to user needs and build trust. Key principles include:
Feedback Loops: Refine AI outputs based on user input.
Flexibility and Personalization: Allow custom prompts and editing.
Anticipation: Use data to predict user needs.
Resilience: Treat errors as opportunities for improvement.
Context-Awareness: Adapt to different user situations.
Adaptive Systems Theory Guides the Interaction Design of AI.
Design Process
I created templates for each AI-generated section within the Navigator, ensuring a cohesive and consistent user experience.
The primary flow allows users to open existing projects, navigate through tabs, and access previously generated versions within each section. This flow ensures users can easily manage and track their project progress.
FLOW A.1: Editing with AI through Custom Prompts: Users can input specific instructions for the AI to follow, which provides a sample before finalizing the changes
FLOW A.2: Editing with AI through Suggested Prompts: Users select from a list of suggested prompts to guide AI edits, with an option to review samples before finalizing.
FLOW A.3: Editing Specific Text with AI: Users highlight and edit specific text sections, with AI suggestions and samples provided before confirming changes.
FLOW A.4: Editing without AI Assistance: Users manually edit text without AI involvement, ensuring they have full control over the content.
Outcome
Reimagine Labs is currently in the development and testing phase of their MVP. The wireframes I created are undergoing iterations based on user feedback. The adaptive system principles applied have set a robust foundation for a user-friendly and flexible AI interaction model.
Lessons Learned
Through this project, I gained valuable insights into the human-computer interaction (HCI) aspects of generative AI. I learned that adaptive systems can provide a valuable framework for designing generative AI products. Additionally, balancing user control with AI automation is essential for building user trust with AI products.
