I’ve always been very interested in language, culture, how we communicate and the different influences our native and adopted cultures have on us.
Several years ago I was introduced to the communication pattern charts from “When Cultures Collide” by Richard Lewis. I love how these provide an easy to understand and relate to visual way of comparing and contrasting different cultural tendencies. While they’re in no way exhaustive, I think they serve a role when used properly as a baseline for having conversations – similar to how I approach and use personality tests. At times, these can manifest or re-affirm unfounded stereotypes but in other cases they’re a wonderful tool for further discussion, reflection, and deeper inquiry.
This got me thinking – what would a Large Language Model-driven AI come up with if I asked it to give me varying communication preferences and societal value across cultures. In theory, the LLM should be able to surface connections that draw both on its pattern recognition as well as existing bodies of work. However, it is also likely to paint a biased picture due to over representation of English language training material and under-representation of Native language material.
With all of this in mind, I started generating a few basic prompts as a fun casual experiment.
This is the output of that 30 minutes of playing around and exploring. As I find the results quite interesting, I decided to share these results even though they’re the result of a super casual experiment.
As I read through the different charts, I then had an additional thought – how do these vary across generations? So, I had ChatGPT further render three additional versions for the US, South Korea and Denmark where I had it elaborated on the prompt but for that specific country and age bracket.
Country Level Charts
China
Colombia
Denmark
Egypt
England
Germany
Ghana
India
Italy
Kenya
Kyrgyzstan
Malaysia
Morocco
Norway
Oman
Poland
Saudi Arabia
Scotland
South Korea
Spain
Sweden
Turkey
United States
Vietnam
Zambia
United States Rendered By Age Brackets
South Korea Rendered By Age Brackets
Denmark Rendered By Age Brackets
I hope you found these as fun and interesting as I did. If you’ve got questions about my approach or want to discuss observations further, you’re welcome to ping me on Threads.