Why This Fellowship Exists

AI products encode social science problems at every layer. Training data is overwhelmingly English. Global South contexts are underrepresented in the datasets and overrepresented in the assumptions. The teams building these systems lack the training to see these as social science problems.

Social scientists are trained in the methods AI teams need: extended observation, contextual inquiry, cross-cultural analysis, the ability to see structural patterns that survey data misses. Yet no structured pathway connects this training to where it is needed in tech.

The Polyfoni Fellowship is built to fill that gap. We give social scientists methods, mentors, and a platform to apply what they already know to AI products that affect the communities they study.

About Polyfoni

Polyfoni was founded to bring social science knowledge into AI development, particularly knowledge from regions and communities that are underrepresented in the systems being built. We study AI products the way anthropologists study any social institution: by asking what assumptions are built in, whose knowledge counts, and what happens when a product designed in one context lands in another.

Our team brings expertise across business anthropology, global health research, AI governance, and qualitative methods. We are based in Oslo and work with researchers and practitioners across Europe, Africa, and Asia.

About the Fellowship

Alongside their research, fellows participate in online masterclasses on AI governance and career transition from academia to industry, monthly group sessions with the full cohort for peer feedback, writing workshops, and guest inputs from AI industry practitioners. Fellows are also encouraged to discuss career opportunities with their mentors and the wider Polyfoni network.

What Fellows Work On

Fellows conduct a cultural audit: an empirical analysis of how a specific AI product encodes assumptions about the context it is deployed in, and where those assumptions break. Each audit follows Polyfoni's Cultural Audit Method: assumption mapping, contextual failure identification, severity assessment, and recommendations.

Polyfoni supervisors propose audit themes aligned with our research priorities. Fellows choose and shape their project in close collaboration with their academic and industry mentors. The output is an 8-to-10-page report written for a product team, published on Polyfoni's Substack under the fellow's name.

The audit target should be a real AI product in a region the fellow has genuine contextual knowledge about. Your fieldwork experience is the data. You are not starting from scratch. You are applying what you already know through a structured method.

Examples of audit targets:

• A health AI chatbot in East Africa

• A content moderation system in Southeast Asia

• An agricultural AI tool in West Africa

• A government AI procurement system in Latin America

• An LLM deployed in a non-English-dominant market

Qualifications and Selection Criteria

We encourage you to apply if you are interested in using your social science training to contribute to how AI products are built, evaluated, and governed.

We are interested in candidates from a broad set of disciplinary backgrounds, including anthropology, sociology, STS, development studies, political science, humanities, and area studies. There are no strict degree requirements, although we expect that the most promising candidates will have fieldwork experience in a Global South context and an interest in technology careers. Prior AI experience is not required.

When assessing applications, we will be looking for:

Regional expertise with evidence of fieldwork: Deep contextual knowledge of a place, community, or sector where AI is being deployed. Priority regions: East Africa, South Asia, Southeast Asia, Latin America, China.

Clear motivation: A specific reason for wanting to work at the intersection of social science and AI, and a concrete idea of which AI product you would audit.

Language: Comfortable working in English. Additional languages are an advantage.

We select for diversity of region and discipline.

What to Expect

• Dual mentorship from an academic collaborator and an AI industry professional

• Training in Polyfoni's Cultural Audit Method

• Online masterclasses and monthly group sessions with the full cohort

• A published cultural audit on Polyfoni's website under your name

• The opportunity to present at the Making AI Listen workshop in March 2027

How to Apply

Application is now open. The selection process has two steps:

Step 1: Initial application. Submit a short application with motivation (500 words maximum). Successful candidates will be contacted for a CV review.

Step 2: Interview. Final candidates will have an interview and a research discussion with the Polyfoni team.

Polyfoni matches fellows with mentors based on regional expertise alignment (your fieldwork region and the mentor's company market) and sector alignment (health, fintech, content moderation, etc.).

We welcome applicants from across the humanities and social sciences, from MSc and PhD students to early-career researchers in adjacent fields. Working on AI governance requires many different perspectives and forms of expertise. If our mission resonates with you, we encourage you to apply, even if your experience does not align with every qualification listed.

Apply using this link. 

Submissions are accepted on a rolling basis. We particularly encourage submissions from early-career researchers and scholars based in the Global South.

If you have questions, contact contact@polyfoni.no