Adam Acunin | Portfolio 2026

ScholarlyAI

led the end-to-end design of an ai chatbot for educators, grounded in user research and validated through testing

UX Research

Usability Testing

Web

OVerview

Because the tool was developed for research, the project followed a fixed schedule and direction. I learned to make thoughtful design decisions within those constraints and advocate for improvements where possible.

TIMELINE

Mar - Aug 2025

ROLE

UX Designer & Researcher

TEAM

3 Researchers, 3 Developers, 1 Teaching Expert

TOOLS

Figma, Slack, Zoom, GitHub

RESULTS AT A GLANCE

This project strengthened my understanding of AI design, usability testing, and feedback-driven iteration.

The findings from this research were published in the following paper, where I contributed as a co-author: Exploring Conversational Design Choices in LLMs for Pedagogical Purposes

experience HIGHLIGHTS

Led 8 usability testing interviews with teachers in higher education

Resolved bugs during QA by pushing code via GitHub

Collaborated with teaching experts to better understand user needs

WHAT IS SCHOLARLYAI?

ScholarlyAI is an AI chatbot designed to support teachers in their educational practice.

This project was completed during my time at MetoyerLab, an HCI lab exploring the intersection of social issues and technology. This research focused on understanding teachers’ experiences and perspectives on AI, with the goal of informing future research and publications.

the question

Students are already using AI in the classroom. But how are teachers using it?

Insight 01

Most AI tools generate direct answers rather than facilitate learning.

How do we design an AI tool that can foster meaningful reflection?

Insight 02

Some teachers experiment actively with AI, while others remain skeptical.

How do we address issues regarding trust, privacy, and transparency?

Insight 03

AI tools for teachers should adapt to their goals, context, experience, and practice.

How do we account for different levels of teaching experience?

problem statement

How might we design an AI tool that supports the diverse needs and behaviors of teachers?

Analyzing user research

To inform ScholarlyAI's design and features, the team conducted 10 user interviews with teaching experts.

These interviews helped us understand teacher's attitudes about AI. Users shared thoughts about transparency, privacy, and brainstorming.

Analyzing user research

To inform ScholarlyAI's design and features, the team conducted 10 user interviews with teaching experts.

These interviews helped us understand teacher's attitudes about AI. Users shared thoughts about transparency, privacy, and brainstorming.

competitive analysis

I analyzed current tools and identified reusable components to create a familiar, low-friction experience.

This process helped me create wireframes for the onboarding, dashboard, and AI chat interfaces, which the team identified as key product requirements.

competitive analysis

I analyzed current tools and identified reusable components to create a familiar, low-friction experience.

This process helped me create wireframes for the onboarding, dashboard, and AI chat interfaces, which the team identified as key product requirements.

Design iterations

I incorporated team feedback to improve the AI chat interface, making referenced resources easier to see.

Stakeholders noted that the initial design obscured chat history. I redesigned the layout to let users quickly reference information across chat history and the resource panel.

Design iterations

I incorporated team feedback to improve the AI chat interface, making referenced resources easier to see.

Stakeholders noted that the initial design obscured chat history. I redesigned the layout to let users quickly reference information across chat history and the resource panel.

cross-functional collaboration

During the development phase, I collaborated with engineers and pushed code to production, streamlining the QA process.

With a tight deadline before usability testing, QA time was limited. I leveraged my HTML/CSS and GitHub knowledge to independently identify and fix bugs, helping the team stay on track.

onboarding design

Personalized onboarding helps tailor responses, while giving skeptical users the freedom to skip it.

The tool gathers user information on years of experience, courses taught, and current views of AI to provide context for future chats.

onboarding design

Personalized onboarding helps tailor responses, while giving skeptical users the freedom to skip it.

The tool gathers user information on years of experience, courses taught, and current views of AI to provide context for future chats.

The Resource Panel

Source materials appear alongside generated responses, providing teachers with greater transparency.

Users can view resources alongside their chat history for easy reference and save resources to their dashboard for future use.

The Resource Panel

Source materials appear alongside generated responses, providing teachers with greater transparency.

Users can view resources alongside their chat history for easy reference and save resources to their dashboard for future use.

building trust with the dashboard design

The dashboard centralizes data controls, teaching resources, and live consultation scheduling in one place.

Users can schedule in-person consultations with the learning center, fostering community discussions about AI. They can also delete their data and access previous resources.

building trust with the dashboard design

The dashboard centralizes data controls, teaching resources, and live consultation scheduling in one place.

Users can schedule in-person consultations with the learning center, fostering community discussions about AI. They can also delete their data and access previous resources.

usability testing

I led 8 usability testing interviews with teachers, comparing a Socratic vs. Narrative version of the tool blindly.

Socratic was more conversational and prompted reflection. Narrative had longer, more detailed responses with examples to deliver information.

usability testing

I led 8 usability testing interviews with teachers, comparing a Socratic vs. Narrative version of the tool blindly.

Socratic was more conversational and prompted reflection. Narrative had longer, more detailed responses with examples to deliver information.

user insights

Teachers with less years of experience preferred the Socratic version, as it provided more guided direction.

Users felt like this version was similar to a consultant helping them think through problems. However, others found this version frustrating when they wanted immediate solutions.

user insights

Teachers with less years of experience preferred the Socratic version, as it provided more guided direction.

Users felt like this version was similar to a consultant helping them think through problems. However, others found this version frustrating when they wanted immediate solutions.

user insights

Teachers with more years of experience preferred the Narrative version, valuing it for its actionable ideas.

Users appreciated having a broad list of possible strategies to implement. However, others described being overwhelmed from information dumping.

user insights

Teachers with more years of experience preferred the Narrative version, valuing it for its actionable ideas.

Users appreciated having a broad list of possible strategies to implement. However, others described being overwhelmed from information dumping.

Design feedback

While users with prior AI experience appreciated the familiarity of the design, they expected quick, low-effort answers.

Users familiar with ChatGPT expected low-effort, option-rich interactions and sometimes resisted interfaces that required deeper reflection, limiting engagement.

Design feedback

While users with prior AI experience appreciated the familiarity of the design, they expected quick, low-effort answers.

Users familiar with ChatGPT expected low-effort, option-rich interactions and sometimes resisted interfaces that required deeper reflection, limiting engagement.

next steps

This research points to more adaptive, personalized chat experiences.

Possible features include toggling between different chat styles or enabling the system to adapt based on user profiles. Designing for AI means designing not only the visual interface, but also how the system responds to users.

reflection & Challenges

Learned the importance of research in rapidly evolving fields like AI

Strengthened interpersonal skills through teacher interviews

Learned to navigate tight timelines and adapt to changing needs

my takeaway

As AI evolves, we can design more intentional AI tools through deeper user understanding.

As AI reshapes how we interact with technology, designers must consider how people understand and trust these systems.

← Previous project

myTickets

NEXT PROJECT →

let's make something beautiful ;)

SEE MY RESUME

aacunin2@nd.edu

let's make something beautiful ;)

SEE MY RESUME

aacunin2@nd.edu

Because the tool was developed for research, the project followed a fixed schedule and direction. I learned to make thoughtful design decisions within those constraints and advocate for improvements where possible.

This project strengthened my understanding of AI design, usability testing, and feedback-driven iteration.

Led 8 usability testing interviews with teachers in higher education

Resolved bugs during QA by pushing code via GitHub

ScholarlyAI is an AI chatbot designed to support teachers in their educational practice.

Students are already using AI in the classroom. But how are teachers using it?

Most AI tools generate direct answers rather than facilitate learning.

Some teachers experiment actively with AI, while others remain skeptical.

AI tools for teachers should adapt to their goals, context, experience, and practice.

How might we design an AI tool that supports the diverse needs and behaviors of teachers?

To inform ScholarlyAI's design and features, the team conducted 10 user interviews with teaching experts.

To inform ScholarlyAI's design and features, the team conducted 10 user interviews with teaching experts.

I analyzed current tools and identified reusable components to create a familiar, low-friction experience.

I analyzed current tools and identified reusable components to create a familiar, low-friction experience.

I incorporated team feedback to improve the AI chat interface, making referenced resources easier to see.

I incorporated team feedback to improve the AI chat interface, making referenced resources easier to see.

During the development phase, I collaborated with engineers and pushed code to production, streamlining the QA process.

Personalized onboarding helps tailor responses, while giving skeptical users the freedom to skip it.

Personalized onboarding helps tailor responses, while giving skeptical users the freedom to skip it.

Source materials appear alongside generated responses, providing teachers with greater transparency.

Source materials appear alongside generated responses, providing teachers with greater transparency.

The dashboard centralizes data controls, teaching resources, and live consultation scheduling in one place.

The dashboard centralizes data controls, teaching resources, and live consultation scheduling in one place.

I led 8 usability testing interviews with teachers, comparing a Socratic vs. Narrative version of the tool blindly.

I led 8 usability testing interviews with teachers, comparing a Socratic vs. Narrative version of the tool blindly.

Teachers with less years of experience preferred the Socratic version, as it provided more guided direction.

Teachers with less years of experience preferred the Socratic version, as it provided more guided direction.

Teachers with more years of experience preferred the Narrative version, valuing it for its actionable ideas.

Teachers with more years of experience preferred the Narrative version, valuing it for its actionable ideas.

While users with prior AI experience appreciated the familiarity of the design, they expected quick, low-effort answers.

While users with prior AI experience appreciated the familiarity of the design, they expected quick, low-effort answers.

This research points to more adaptive, personalized chat experiences.

Learned the importance of research in rapidly evolving fields like AI

Strengthened interpersonal skills through teacher interviews

Learned to navigate tight timelines and adapt to changing needs

As AI evolves, we can design more intentional AI tools through deeper user understanding.

myTickets

myTickets