AI-Mediated Communication in the Workplace
An AI-based plug-in to assist in workplace or professional communications.
Project Overview
This was a group project carried out in my "Future of UX and AI" class.
We were given the following prompt: "How might the increasing dependence on AI in daily life impact human interpersonal relationships and social skills, and what measures can be implemented to ensure a balanced coexistence?"

To narrow down our research, we focused on the impact of AI-mediated communication within the domain of professional communication.
What I did
  • Primary Research
  • UX/UI Design
  • Interaction Design
  • September 2023 - December 2023
  • 3 Months

Project Brief
  • “[AI-Mediated communication is the] Interpersonal communication optimized, augmented, or even generated by algorithms to achieve specific communicative or relational outcomes.” — Jeff Hancock
  • There are several kinds of AI implementations in communication, for example, voice-assisted communication, language correction, predictive text, translation, and transcription.

  • Our ultimate research question was:
    “Does Artificial Intelligence (AI) influence people’s communication skills and relationships in the professional setting, specifically for product managers, designers, and engineers? If so, in what ways and why?”

Research Objectives
  • Identify professionals' needs and challenges when employing AI for composing messages.
  • Determine the strategies they adopt to integrate these tools into their workflows.
  • Examine how recipients perceive AI-generated messages' authenticity, clarity, and efficacy within professional communications.

Research Methods
  • Survey Form
  • 1-on-1 Stakeholder Interviews with:
    • → Software Engineers
    • → UX Designers
    • → UX Researchers
    • → Product Managers
  • Secondary Research
  • User Testing

Research Findings
We collected the following data through survey forms and interviews. We carried out 1-on-1 interviews with 13 participants.
Our Insights
  • Adjusting the tone of a message was the most popular task or need.
  • The tones in AI-generated responses often fail to match with user’s personalities or contexts.
  • Many interviewees struggle to find the right tone, posing a challenge for timely responses.
  • Uncertainty about expressing results leads to hesitancy in reaching out.
  • Current Challenges with AI-Generated Texts

Current Challenges with AI-Generated Texts
  • Unsatisfactory responses are due to bad prompts
  • Ineffective communication
  • Not enough experience at a new job or environment
  • Overdependence on AI

Our Design Solutions

We designed a plug-in, called ChatAX, to integrate with Gmail for assisting in email drafting. This will have 3 features: smart reply, customize message, and detect tones.

Smart Reply will show the user suggestions for a reply. This will be similar to predictive text suggestions. The user can also enter a specific message tone to help with the suggestions.

Customize Message feature lists out available options and provides the space to describe any specific needs.

Then, the user can select sender and receiver's job titles to find the best tone for the message.

User can also pick from a wide range of tones and languages.

At the end, the user can provide more context to help ChatAX draft the most appropriate message.

The last feature detects tones in already drafted messages. It also provides suggestions to improve the drafts.

Finally, the user can make any edits they want before copying the AI-generated message. They can even translate the message into different languages.

User Testing

Among the 5 professionals we conducted user testing with:

  • All participants are able to use the plugin without assistance.
  • All participants think referring to a list of tones will help them communicate more accurately.
  • Users need a detailed hierarchy for position searches.
  • More context will help AI find the right tone.

Based on the user feedback, we improved two elements in our first design:

Final Prototype

Through this project, I got to learn the limitations of AI and explore the appropriate integration of AI in a user's work and life. I interviewed users and learned about their needs, goals, and fears linked to AI technologies. So, this project pushed me into the realm of AI ethics and how I can work to make sure new AI-based solutions do not violate the users' privacy and safety.