Why Intercultural Competence Matters More in the Age of AI
AI can produce fluent answers across cultures, but understanding them still depends on a developed human capacity
Does Anyone Still Need Intercultural Competence When AI Can Translate Anything?
Yes, and probably more than before. It is a fair question to ask. A student can paste a paragraph into a translation tool and read it in seconds. A staff member can ask an assistant to summarize how another country approaches higher education policy. A manager can draft a message to a colleague overseas and have it polished before lunch. When tools handle that much, it is natural to wonder what is left for people to learn.
What is left is the part the tools do not do for you. Translation moves words from one language to another. It does not tell you why a colleague stayed quiet in a meeting, what a pause means in a particular setting, or how your own assumptions shaped the way you read a message. Those are questions of interpretation, and they stay with the person.
Fluent Output Is Not the Same as Cultural Understanding
Intercultural competence is the capability to shift cultural perspective and appropriately adapt behavior to cultural differences and commonalities. It is a developmental capacity rather than a store of facts. A person can know a great deal about a culture and still struggle in a real interaction with someone from it. We look at that distinction in more detail in our explainer on what intercultural competence is and why it matters on campus.
AI sharpens the distinction. These tools are very good at producing confident, well organized information about cultures, and information is not the same as competence. The Intercultural Development Continuum describes how people make sense of cultural difference, from the more monocultural orientations of Denial and Polarization, through the transitional orientation of Minimization, to the intercultural orientations of Acceptance and Adaptation. Where a person sits on that continuum shapes what they do with information once they have it.
IDI research points the same way. It finds that interculturally competent behavior occurs at a level supported by a person's underlying orientation, and you can explore that body of work in the IDI's Research and Publications library. A polished answer from a tool does not lift someone beyond where they currently are.
What the Research Says About Cultural Bias in AI Tools
The more closely researchers look at how AI tools handle culture, the clearer it becomes that the output is not culturally neutral. A study published in PNAS Nexus by a Cornell led team compared the responses of five widely used OpenAI models with nationally representative survey data from countries around the world. All five models produced values that resembled those of English speaking and Protestant European countries. The researchers also tested a simple fix, which they call cultural prompting, where the user specifies a cultural perspective in the request. For the newer models, it improved alignment for roughly 71 to 81 percent of countries and territories. You can read the abstract in Cultural Bias and Cultural Alignment of Large Language Models.
A separate preprint looked at moral reasoning across 19 cultural contexts. It reports that the models it tested tended to flatten moral diversity, and that making a model larger did not consistently fix the problem. That work has not yet been through peer review, so it is best read as an early signal. You can find it as Cultural Bias in Large Language Models: Evaluating AI Agents through Moral Questionnaires.
It is worth keeping this in proportion. Results vary by model and by task, and the tools are changing quickly. The consistent finding is not that AI is useless across cultures. It is that the answers carry a perspective, and that perspective is not evenly distributed around the world.
How AI Can Quietly Reshape the People Who Use It
Bias in a tool matters most when it changes what people do. Researchers at Cornell studied exactly that in an experiment with 118 participants from India and the United States. Participants completed culturally grounded writing tasks, some with AI writing suggestions and some without. The tool produced larger efficiency gains for the American participants. Indian participants who used it began to write in a more Western style, and in some cases described their own food and festivals through a Western lens. The study was presented at the ACM CHI conference and is available as AI Suggestions Homogenize Writing Toward Western Styles and Diminish Cultural Nuances.
One detail stands out. The Indian participants accepted slightly more of the suggestions than the Americans did, but they also edited them more heavily to fit their topic and voice. In other words, they were doing extra work to keep their own perspective in the writing. A person who does not notice that pull, or does not have the awareness to push back, may never realize their expression has shifted.
Why Orientation Toward Difference Shapes How People Use AI
The studies above describe the tools and their effects. They do not test how a person's intercultural development changes the way they use AI, so what follows is a reasonable implication rather than a finding.
Consider a reader who tends toward Minimization, where similarities are emphasized in ways that can hide meaningful differences. A fluent answer that treats one cultural norm as universal is likely to feel natural and correct. A reader in Denial or Polarization may use confident output to confirm a view they already hold. A reader oriented toward Acceptance or Adaptation is more likely to ask whose perspective is missing, to check an answer with someone who has lived experience, and to adjust the request. The tool is the same in each case. The use is different, and the difference comes from the person.
This is why the Intercultural Development Continuum is useful beyond the classroom. It gives people a way to see their own starting point and to build the habits that make AI output something to question and refine instead of something to accept.
What This Means for Campuses
Colleges and universities are where many people first use AI for work that crosses cultural lines. Students use it to prepare for study abroad, to draft assignments, and to get feedback on writing. Advisors and staff use it to communicate with international students and families. Faculty use it to build materials for classrooms that include students from many backgrounds. Each of those uses carries the risks described above, and each is also an opportunity to teach something.
A few practical steps follow. Teach students to treat AI answers about cultures as a starting point rather than an authority. Encourage them to ask whose perspective an answer reflects and to name a cultural context in their requests. Pair AI use with conversation with real people, especially in study abroad preparation, where a tool can supply facts but cannot replace guided reflection. We make the case for that pairing in our piece on whether study abroad alone builds cultural competence.
There is also a measurement point. Tools make it easier than ever to look culturally aware in writing and communication without having developed the capacity underneath. Programs that want to know whether students are actually growing need a way to see orientation, not polish. That is the role pre and post assessment plays, and our post on how higher education programs measure the impact of intercultural learning shows what that looks like in practice. Because the IDI is administered by trained Qualified Administrators, campuses can build that capacity internally or work with one through the QA database.
Where to Start
None of this requires a campus to take a position for or against AI. The tools are here, and people will keep using them. The more useful question is whether the people using them have the awareness to notice what a tool is doing with culture and the skill to respond. That awareness is not built by the technology. It is built through reflection, practice, and a clear picture of where a person currently stands.
Intercultural competence was always about how people make sense of difference and adapt to it. AI does not change that. It raises the stakes, because it puts culturally flavored output in front of more people, more often, with more confidence than ever before. The people best prepared for that are the ones who have developed the capacity to question it. To learn more about the framework behind that development, start with the Intercultural Development Continuum.
Related Reading
Continue with our explainer on what intercultural competence is, our look at whether study abroad builds cultural competence, and our guide to measuring the impact of intercultural learning.
References
Agarwal, D., Naaman, M., & Vashistha, A. (2025). AI suggestions homogenize writing toward Western styles and diminish cultural nuances. In CHI Conference on Human Factors in Computing Systems (CHI '25). Association for Computing Machinery. https://doi.org/10.1145/3706598.3713564
Münker, S. (2025). Cultural bias in large language models: Evaluating AI agents through moral questionnaires [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2507.10073
Tao, Y., Viberg, O., Baker, R. S., & Kizilcec, R. F. (2024). Cultural bias and cultural alignment of large language models. PNAS Nexus, 3(9), Article pgae346. https://doi.org/10.1093/pnasnexus/pgae346
Frequently Asked Questions
Can AI replace intercultural competence?
No. AI can supply information and translation, but intercultural competence is a person's capacity to shift perspective and adapt behavior in real interactions. A tool does not develop that capacity for the people who use it.
Is AI culturally biased?
Research suggests that many widely used models produce answers closest to the values of English speaking and Protestant European countries. Results vary by model and task, and the tools are changing, so the safest approach is to treat output as one perspective rather than a neutral one.
How can people use AI more thoughtfully across cultures?
Treat answers as a starting point, ask whose perspective the answer reflects, name the cultural context in your request, and check important conclusions with people who have lived experience of that context.
Does the IDI measure how people use AI?
No. The IDI measures how a person or group makes sense of and responds to cultural differences. It does not assess AI skills, but a person's orientation toward difference can shape how they interpret and use any tool.
Why does assessment matter more now that AI can polish communication?
Because AI can make writing and messaging look culturally aware without any underlying development. Assessment shows a person's orientation toward difference, which polished output cannot reveal.