MIT Media Lab Portfolio

Hi, I’m Karina Bhattacharya. I am applying to the Media Arts and Sciences program at MIT Media Lab for Fall 2026. I am particularly interested in the following groups: Personal Robots, Cyborg Psychology (Cyborg Interface Designer and Cyborg Anthropologist), and Multisensory Intelligence.

Please navigate this page on my website for my related portfolio projects. Note that this page’s link is unlisted (no password required) and can be accessed directly via the link on my application.

---

The following projects are hyperlinked below:

Human and AI Biases in Participatory Research
1.1 Case Study in Teaching
1.2. Digital Tool for Checking Biases from Humans and AI

2. AI as a Visual Thinking Collaborator in Co-Design

3. Co-Design Study with Neurodivergent Students

---

1. Human and AI Biases in Participatory Research

Case Study in Teaching

My Role: Faculty Instructor in Industrial Design
SummerFAB Pre-College Program at Wentworth Institute of Technology
Week 1 of the 4-Week Program

Interview Prep

As a faculty instructor, I introduced the high school students to ethnographic research. I organized the students into 4 groups, each directed by a TA who served as a facilitator to develop questions for the community. To ensure there was minimal overlap between the different groups’ questions, I designated question categories:

• About the location
• About the people
• Actions taking place
• How people feel in the space
• How they’d like to represent themselves

/

Synthesizing as a Group

We all reconvened in the studio and the groups wrote pieces of evidence on different pages. They organized the pages on the wall, and I facilitated a group discussion.

I typed up each piece of evidence into a document and uploaded it to ChatGPT. I provided context about our program’s mission and prompted it to synthesize our findings, identify our human biases, and self-acknowledge its biases as an AI.

Points Discussed with the Students

• False Statements: ChatGPT stated that the garden is “currently maintained mostly by elderly Latinx women,” which we did not mention in our collection of evidence.
• In-Person Nuances: ChatGPT acknowledged that it “cannot access nuances of tone, body language, or unspoken dynamics,” which caused us to consider human-to-human dynamics as critical evidence of our ethnographic research.

Conclusion

In the early stages of SummerFAB, I facilitated the high school students’ first exposure to in-person ethnographic research. They learned the basics of preparing an interview guide, communicating with community members, and conducting observations on-site.

The AI bias-checking exercise demonstrated that both human researchers and AI introduce biases into ethnographic research. However, we can use AI to synthesize findings while maintaining a critical mindset about its limitations.

Digital Tool for Checking Biases from Humans and AI

Work In Progress

I am currently building a custom GPT for researchers to establish context and upload their data (i.e. interview transcripts and field notes) to check for potential biases by humans and AI. Check back here for updates!

---

2. AI as a Visual Thinking Collaborator in Co-Design

Independent Researcher
Industrial Design Pathways Project
November 2024 – Present

I conducted a virtual co-design study with seven recent graduates who majored in industrial design but pursued design-adjacent careers post-graduation.

A goal of the project was to co-design a website for current students to learn about various career pathways they might be interested in.

1. We gathered inspiration images to define the theme.
2. We prompted Generative AI with an idea for the website.
3. They evaluated the AI images and chose 1 to define the concept.

---

---

Organizing the Co-Designed Concepts and AI-Generated Images

Conclusion

To co-design a career resources website, we used Generative AI as a visual thinking collaborator. We provided inspiration images to define the theme, explained our website idea, and prompted the AI to generate a set of conceptual photos.

My co-designer then chose one AI-generated image to clarify their vision. I would like to explore further how AI can be optimized as a catalyst for visually articulating and refining human creative intentions in collaborative settings.

---

3. Co-Design Study with Neurodivergent Students

Co-Design Workshop Format (First Draft)

A first draft of the workshop format was tested in a pilot study of the co-design session. I had wanted to see if the neurodivergent designers could ideate different ways of approaching the design discovery and ideation processes.

However, this exercise was noticeably difficult for the pilot study’s test participants. Perhaps these designers did not have a unique approach to these processes, which prompted me to later develop open-ended prompts for co-design.

Co-Design Workshop Format (Version 2)

“Overview for Today” The workshop had a clear structure and allotted breaks to accommodate the anticipated needs of the neurodivergent co-designers.

1. “Learning more about you” prompted the co-designers to reflect upon their design education and the project-related habits they had previously developed.

2. “Co-Design” encouraged the co-designers to reflect upon the current and future states of design education and create actionable solutions.

3. “Reflection” gave the co-designers the opportunity to evaluate their work and state pros, cons, and next steps. This provided insights on how their solutions could be implemented in the future.

Analysis of the Workshops

I used thematic analysis (TA) to analyze the qualitative data from the interview transcripts and notes.

1. Responses from the interview transcripts were filtered through as Suggestions or Success Criteria.
2. Suggestions were organized into 3 types of educational interventions. The most notes were in instructional delivery.
3. The resulting data were grouped by relevance to 6 concept keywords.
4. To illustrate what the co-designers meant, the keywords were mapped to direct quotes from the interviews.

---

See the project page to learn more about the outcomes of this research.