Cultural Adaptation
Cultural adaptation was not a cosmetic concern for Samasource but a structural operational requirement. The organization's 2013 expansion to India revealed that systems built in East Africa — training programs, quality frameworks, feedback mechanisms, management practices — did not transfer to Kolkata without significant modification. The failure was not in the workers, who were equally capable; it was in the embedded cultural assumptions the systems had not examined because they had never needed to. Feedback delivered directly to individual workers, which worked in Nairobi's professional context, was received in Kolkata as public criticism carrying implications about competence and status. The feedback was technically identical. The cultural reception was categorically different. Resolving the mismatch required developing a culturally adapted approach — feedback through team channels, positive examples emphasized, quality framed as collective progress — and months of institutional attention to recognize that what had looked universal was actually East African.
In the AI Story
The principle has direct and uncomfortable implications for the AI transition. The large language models powering tools like Claude Code are products of a specific cultural context — American technology culture with its particular assumptions about direct communication, iterative specification, and flat professional interaction. These assumptions are embedded in the tools at every level, not just in the English-language interface but in the deeper logic of how the tools interact with users.
A developer whose professional communication norms emphasize contextual framing, narrative explanation, and relational establishment before task specification may find AI tools less productive than a developer whose communication style aligns with the tool's expectations. The difference is not intelligence or competence. It is cultural fit — the degree of alignment between user communication patterns and the patterns the tool was optimized for.
The attribution error is the additional cruelty. A developer in San Francisco finds AI tools intuitive and attributes the productivity to tool quality. A developer in Lagos finds the same tools less intuitive and attributes the gap to her own deficiency. The cultural bias in the tool's design produces divergent trajectories that widen the gap between culturally aligned and culturally misaligned users — a gap that appears to reflect differences in talent but actually reflects differences in institutional accommodation.
The technology industry's standard approach to cultural adaptation is localization — translating interface into local languages, adjusting date formats and currency symbols, perhaps modifying imagery. Localization addresses the surface of cultural difference. It does not address the depth. Genuine cultural adaptation of AI tools would require engagement with the communication patterns, professional norms, and workflow conventions of the cultures in which the tools are deployed — engagement at the level of the interaction model, not just the interface. The investment is expensive and institutionally demanding, and it has not been made at the scale the global deployment of AI tools would require.
Origin
Cultural adaptation crystallized as an operational principle through Samasource's 2013 India expansion, which produced specific documented failures that the organization had not anticipated and that reshaped its subsequent approach to every geographic expansion.
Janah treated the resulting lessons as transferable to any technology deployment across cultural contexts, and she articulated them in her 2018 writings and in the penultimate chapters of Give Work.
Key Ideas
Universality is cultural. The claim that a system works 'everywhere' is made by people who have tested it in their own context and assumed the results would generalize — an assumption that reflects cultural confidence rather than actual universality.
Depth beyond localization. Genuine cultural adaptation requires changes at the level of the interaction model, workflow conventions, and professional norms — not merely translation of interface language and cosmetic adjustment of display conventions.
Attribution asymmetry. Culturally aligned users attribute their productivity to tool quality; culturally misaligned users attribute the gap to their own deficiency. The error compounds over time into divergent capability trajectories.
Institutional investment. Cultural adaptation requires sustained engagement between tool designers and tool users from different contexts — the kind of investment technology companies have historically been reluctant to make at the scale global deployment would require.
Appears in the Orange Pill Cycle
Further reading
- Leila Janah, Give Work, Penguin, 2017.
- C.K. Prahalad and Venkat Ramaswamy, The Future of Competition, HBS Press, 2004.
- Geert Hofstede, Culture's Consequences, Sage, 1980.
- Erin Meyer, The Culture Map, PublicAffairs, 2014.