Is AI Culturally Intelligent?

The days of human messaging are numbered. Website chat boxes, which live customer service representatives once staffed, are now powered by Artificial Intelligence (AI). Even the emails we write ourselves are becoming more the words of AI. According to Menio Ventures, 1 in 5 adults are now using AI to write their emails. The growing reliance on AI reflects a shift in how individuals and organizations approach communication.

While large language models (LLMs) are generating the content, the audience consuming it remains human. This shift raises important questions: How do we maintain authenticity, empathy, and cultural relevance in a world where machines are doing the talking? How do we ensure that communication, no matter whose voice, continues to build trust and connection?

Today, many models present themselves as all-knowing and adaptable to global needs. LLMs are adding more human characteristics and trying to make their appearance more personable to users. We see the messaging they write for us and may be amazed at how polished it reads. But how do global audiences perceive that messaging? How well do these models truly understand and adapt to global differences?

Which Humans?

A team at the Department of Human Evolutionary Biology at Harvard University recently attempted to answer this question by testing ChatGPT for their paper, “Which Humans?“ The findings revealed that the model tends to think and respond in ways that align more closely with individuals from North America and Western Europe than with those from Africa, East Asia, and South America.

AI is not a neutral force; it does not adapt to the values and preferences of its users and the world. It reflects the perspectives of the people and cultures that influenced it, which are frequently Western-oriented. Relying on these models to navigate global interactions and planning is unlikely to be successful, as it cannot yet relate to and adapt to all humans.

Cultural Values vs. Behavioral Preferences

The biggest area where LLMs are going to struggle in terms of global flexibility is their models’ ability to understand and respond to different cultural values and behavioral preferences.

Cultural values are the shared beliefs and ideals within a society that define what is good, right, and important. They help guide behavior, decision-making, and social norms.

Behavioral Preferences are the ways people prefer to navigate life as individuals.  These values indicate how people prefer to interact, communicate, plan, and complete tasks.

Cultural Values are how a society influences its people and their day-to-day lives. If LLMs only align with their designers’ cultural value systems, they will not come across as relatable globally but only in regions with shared values.

However, as the models become more personable, they must also adapt to our individual behavioral preferences. From how we communicate to how we lead, our natural tendencies influence the way we interact with the world. Recognizing these behavioral preferences not only in ourselves but also in others opens the door to stronger relationships, better teamwork, and more effective leadership.

AI and LLMs are lacking the ability to adapt to the cultural values and behavioral preferences of their users. The skill that AI is missing is Cultural Intelligence. Cultural Intelligence, or CQ, refers to the ability to relate to and work effectively with people from diverse cultural backgrounds, extending beyond traditional notions of cultural sensitivity and awareness. Individuals with High CQ can separate themselves from AI and make a greater impact across cultures. Try a CQ assessment to see where your CQ is and gain actionable insights on how to develop it.