Designing AI systems that travel well across cultures, regions, and the Global South — with fairness, accountability, and context.
Across the world, AI systems influence who is prosecuted, who is fired, and who receives medical treatment. Yet many of these systems are conceived for Silicon Valley contexts. They often fail to account for the realities of the Global South or rural America, constrain local creativity, and harm communities they were not designed for.
As designers and builders of AI, can we imagine systems that travel well across cultures and regions? How do we anticipate bias before deployment, and how do we guard against unintended consequences? What does equitable, accountable, and culturally grounded AI look like in practice?
In this seminar, students survey the research landscape on globally aware AI with a focus on interface and system design. Each week we read and discuss work on cross-cultural design, fairness and accountability, participatory and community-centered methods, and case studies from multiple regions. Students complete a research project that builds, evaluates, or theorizes a design of an AI system aimed at real-world use beyond the West.
Lectures include:
Upon completion of this course, you will:
This course treats culture not as a fixed checklist of national traits, but as a dynamic system of values, histories, institutions, languages, identities, and everyday practices. We will examine how AI systems reflect cultural assumptions, how they travel across contexts, and how they can be redesigned to better support diverse communities.
Students will learn through seminar discussion, design critique, hands-on prototyping, case analysis, and reflective writing. The course combines critical reading with practical design work so that students can both analyze AI systems and imagine alternatives.
Learning experiences include:
By the end of the course, students will have practiced both critical and applied skills for designing culturally aware AI systems.
Students will develop skills in:
The final project serves as a synthesis of the course. Students will identify a cultural challenge related to AI, investigate the needs and values of relevant stakeholders, and develop a project that proposes, evaluates, or theorizes a culturally aware AI system.
Projects may take several forms, including:
Students will be expected to justify their design choices using course readings, cultural analysis, stakeholder needs, and evidence from their own research or evaluation.
Want to teach this class at your own university? Please reuse, adapt, and share the slides, assignments, and course materials. Feedback on how the course works in your context is most welcome — please reach out by email.
Course overview and objectives, what it means to think and work globally, understanding different types of work patterns for designing interfaces for different cultures, and examples of global work.
Submit a one-page summary including:
Readings:
Videos:
Answer the following questions by drawing direct connections between the texts and the videos:
This week covers the core concepts of culture and user behavior; the intersection of human culture and AI; real-world AI failures driven by cultural blindness; why cultural context is critical for AI design; and an introduction to the Layers of Culture framework.
AI systems are often designed using assumptions from the Global North — stable connectivity, individualistic norms, and standardized language. This project asks you to redesign an AI product for Nigeria using the Layers of Culture framework.
Company: Apple · Product: Project Sight, AI-integrated smart glasses · Capabilities: real-time audio feedback, translation, navigation, task assistance · Target context: Nigeria (Lagos or Abuja)
Evaluate the role and limitations of Western design frameworks when applied to diverse global populations.
This week covers how culture operates across different levels; Pace Layering and how systems change at different speeds; interactions between organizational cultures, national contexts, and technologies; and how cultural context informs inclusive and effective interface design.
Submit a one-page summary of your final project idea — a culturally aware AI tool that addresses a real-world problem while accounting for cultural context, language, norms, infrastructure, and values.
Your one-pager should include:
Format: Max 1 page, PDF preferred. Teams may submit the same one-pager; each member submits individually.
This week covers what designers need to know about culture to create effective AI systems; comparing cultures across key dimensions; models of culture in HCI and AI research; Hofstede's six dimensions (Power Distance, Individualism vs. Collectivism, Uncertainty Avoidance, Masculinity vs. Femininity, Long-Term Orientation, Indulgence vs. Restraint); and design exercises for different cultural contexts.
Prepare a short 3-minute presentation giving an update on your final project. Explain what you will study, how you will study it, and how you plan to design for it. All team members must submit individually.
Build a Cultural Logic Controller — a Python script that allows an AI system to dynamically shift its personality and decision-making priorities based on the cultural dimensions of its user.
Use Hofstede's Dimensions to decide behavioral strategy:
| Country | Power Distance | Individualism | Uncertainty Avoidance | Long-Term Orientation |
|---|---|---|---|---|
| United States | 40 | 91 | 46 | 26 |
| Germany | 35 | 67 | 65 | 83 |
| Brazil | 69 | 38 | 76 | 44 |
| Vietnam | 70 | 20 | 30 | 80 |
class CultureBot:
def __init__(self, country, power_distance, individualism,
uncertainty_avoidance, long_term_orientation):
self.country = country
self.power_distance = power_distance
self.individualism = individualism
self.uncertainty_avoidance = uncertainty_avoidance
self.long_term_orientation = long_term_orientation
def generate_response(self, task):
# 1. SET THE TONE (Power Distance Index)
# 2. SET THE INCENTIVE (Individualism vs. Collectivism)
# 3. SET THE PRECISION (Uncertainty Avoidance Index)
# 4. WILDCARD: THE TIMELINE (Long-Term Orientation)
pass
usa_bot = CultureBot("United States", 40, 91, 46, 26)
brazil_bot = CultureBot("Brazil", 69, 38, 76, 44)
print(usa_bot.generate_response("Save $100"))
print(brazil_bot.generate_response("Save $100"))
| Deliverable | Weight |
|---|---|
| Python script (.py) | 70% |
| Reflection document (three questions above) | 30% |
| Bonus: Add a fifth country with justified Hofstede scores | +10 pts |
Examining websites, mobile apps, and AI interfaces designed for different countries; identifying how cultural dimensions may have influenced interface features and interaction styles; and using Hofstede's framework as an analytical tool to critique user interface designs.
For each reading and the video, provide:
| Source | Points |
|---|---|
| Reading 1 | 25 |
| Reading 2 | 25 |
| Reading 3 | 25 |
| Video | 25 |
| Total | 100 |
Design intelligent goggles that help U.S. military members have more effective interactions with civilians in foreign countries (Russia, Ukraine, Afghanistan, or other distinct cultural contexts). Communicate your design through a short article.
Format: 2 pages, single-spaced, PDF preferred. Include diagrams, mockups, or storyboards.
Reference: Steed (2009) — Cultural Challenges Facing the Military
Weight: 30% of Final Grade
Your group will act as Seminar Leaders for half a class session (~1 hour). Move beyond presenting information — teach your classmates about a novel intersection between culture and AI. Curate learning materials, deliver a lecture, and facilitate an engaging classroom activity.
A. Preparatory Packet
B. Seminar Presentation
C. Active Learning Activity
Group project. Each student submits the Learning Package individually. One team member submits the package to the TA and instructor on the Friday before the Monday seminar.
After this week you should be able to: describe the Standard Machine Learning Pattern; determine where it fails when connected to culture; describe often overlooked factors that impact the pattern; define common ML performance terminology; justify where iteration is most important; and begin to understand how to create ML models and AI interfaces for different regions.
Read all three papers, then for each provide:
Each reading is worth 30 points (90 pts total). At the end, write 1–2 sentences about what you enjoyed most from the readings (10 pts). Total: 100 points.