Project

Ask UXR

Ask UXR

Faster Research Conclusions

while empowering the designer.

Faster Research Conclusions

while empowering the designer.

Info

Role

UX and AI Designer

Timeline

January 2024-May 2024

Tools/Lanuguages

Figma

Python

Streamlit

Overview

Problem

Researchers have a hard time synthesizing and analyzing large amounts of qualitative data, leading to time-consuming and labor-intensive processes.

Researchers have a hard time synthesizing and analyzing large amounts of qualitative data, leading to time-consuming and labor-intensive processes.

Process

The team created a canvas-style interface that combined chatbot functionality with an iterative loop between the AI and the human.

The team created a canvas-style interface that combined chatbot functionality with an iterative loop between the AI and the human.

Product

The research platform integrated responsible GenAI with an innovative interface, allowing UX researchers to sort through their data in minutes.

The research platform integrated responsible GenAI with an innovative interface, allowing UX researchers to sort through their data in minutes.

The AI x Human System

The Iterative Human and AI Loop

The Iterative Human and AI Loop

The first step of the project was to understand what the system between the human and AI needed to look like. The following diagram showcases a feedback loop that empowers the researchers to create and edit findings while incorporating AI as an efficient and helpful tool.

The first step of the project was to understand what the system between the human and AI needed to look like. The following diagram showcases a feedback loop that empowers the researchers to create and edit findings while incorporating AI as an efficient and helpful tool.

Applying the Loop

We analyzed the individual process in one research step using the Observation stage as an example. The AI analyzes and synthesizes data into a card, a process that can be tedious and time-consuming.


The feedback loop is then integrated into four distinct actions, leveraging AI and human input to empower users. This process is part of a larger research methodology with the goal of continually refining data into actionable insights. With this interface, researchers can track progress, save and build upon their work, and iterate until satisfied with the results.

We analyzed the individual process in one research step using the Observation stage as an example. The AI analyzes and synthesizes data into a card, a process that can be tedious and time-consuming.


The feedback loop is then integrated into four distinct actions, leveraging AI and human input to empower users. This process is part of a larger research methodology with the goal of continually refining data into actionable insights. With this interface, researchers can track progress, save and build upon their work, and iterate until satisfied with the results.

The Interface

Visualizing the Research Workspace

Visualizing the Research Workspace

The canvas-style UI was created to layout a collaborative environment between the researcher and AI.


The interface is divided into two distinct sections: the conversation space and the assets space. The assets space contains AI-generated cards, which are created from previous data or cards and serve as the foundation for user actions.

The canvas-style UI was created to layout a collaborative environment between the researcher and AI.


The interface is divided into two distinct sections: the conversation space and the assets space. The assets space contains AI-generated cards, which are created from previous data or cards and serve as the foundation for user actions.

Product

Final application designs

Final application designs

The final design includes a chatbot interface on the left, which is accompanied by a list of all cards. Each card offers three options: converse with the AI, edit manually, or approve. Once all cards are edited and approved, they are further synthesized for the next process, generated, and added into the next list of cards for the user to parse through.

The final design includes a chatbot interface on the left, which is accompanied by a list of all cards. Each card offers three options: converse with the AI, edit manually, or approve. Once all cards are edited and approved, they are further synthesized for the next process, generated, and added into the next list of cards for the user to parse through.

Intuitive Actions

Intuitive Actions

Intuitive Actions

Action Buttons

The four actions are designed for seamless navigation through AI-generated cards, with easy-access tabs at the top to further categorize and organize the content.

The four actions are designed for seamless navigation through AI-generated cards, with easy-access tabs at the top to further categorize and organize the content.

Editing Through Conversation

Editing Through Conversation

Editing Through Conversation

Chatbot Interface

The interface enables users to submit a card and initiate a conversation about it, with the card's layout integrated into the chatbot to facilitate exploration and collaboration with the AI.

The interface enables users to submit a card and initiate a conversation about it, with the card's layout integrated into the chatbot to facilitate exploration and collaboration with the AI.

Streamlined Methodology

Streamlined Methodology

Streamlined Methodology

Tabular Exploration

The tab navigator on both the Asset section and top nav allows users to peruse through the cards and track their progress and work in each tab.

The tab navigator on both the Asset section and top nav allows users to peruse through the cards and track their progress and work in each tab.

Impact

Informed Future Product Direction

The UXR Chatbot prototype established critical design principles and technical foundations that shaped IBM Software's approach to AI-assisted research tools.

The UXR Chatbot prototype established critical design principles and technical foundations that shaped IBM Software's approach to AI-assisted research tools.

01

Process Innovation

The project's innovative canvas-chat hybrid interface model proved that AI assistance doesn't require sacrificing user agency. In this interface, researchers are enabled with rapid iteration through the uses of AI while accelerating validation cycles, therefore empowering the user and making them more efficient.

02

User Empowerment

When tested with 5 IBm UX researchers who were unfamiliar with the product, 4 out of 5 expressed interest in incorporating the tool into their flow. This validated the balance between AI and human creative control.

03

Cross-Functional Value

The product drives alignment between design and engineering by creating high-fidelity prototypes that were backed up by testing done with Watson-X. This allowed the team to test technical contraints alongside user experience considerations in real-time.

Want to see more?

next project.