22/09/2026
AI can analyse a culture faster than any team by processing its language, images, and feeds to provide a fluent account of its apparent values. However, this account often omits crucial elements.
Cultural meaning systems are shaped as much by what is unspoken as by what is expressed—by codes so ingrained that insiders never write them down, and by those kept off-record. Especially in non-Western cultures, these missing parts are precisely what the training data lacks. The model doesn’t recognise their absence and fills the gaps based on the culture it was trained on, presenting its findings with the same confidence as other outputs.
This overconfidence poses a risk, as misinterpretations can appear polished and convincing early on, influencing decisions before dissenting views emerge. While AI can identify what a culture explicitly communicates, it still can’t discern what it leaves unsaid—often the critical factor behind many failures.