A recent Fast Company article touched on a key shortcoming of artificial intelligence (AI) innovations: These technologies are typically trained, intentionally or not, to adopt the social preconceptions, norms, and assumptions of the environment in which they were created.

Looking specifically at the autonomous vehicle (AV) market, the article explored the different “traffic cultures” that exist around the world, and what companies operating in the space must be conscious of as AI adoption increases. Describing a recent rush hour experience in New Delhi, replete with cars, buses, rickshaws, and “the occasional cow,” the author, Jasper Dekker, argued that an AV trained in Silicon Valley would be ill-equipped to address this flurry of activity.

He wrote, “Beyond the official rules of the road, social and cultural norms define the way we actually drive, and how we treat each other on the road…Without knowledge of the traffic culture it finds itself in, an AV would be paralyzed. If engineers and designers fail to find a way to help AI understand the different traffic cultures, we run the risk of excluding a large part of the planet’s population from a new frontier in transportation.”

So, what is the best path forward?

In order for AVs to truly be ready to deploy globally, the article suggests that AI must be trained locally via hyper-local feedback loops to identify unique patterns, such as the fact that drivers typically yield at rotaries or that they are expected to share the road with animals.

In addition to the obvious safety benefits of ensuring AVs are familiar with local driving customs, it’s also likely that hyper-local AI could help people become more comfortable with the concept of AVs. As the Fast Company article put it, “Local training could potentially decrease the risk of conflicts between human drivers and AVs, and it could make the AV experience feel more familiar to passengers.”

Of course, the industry still has a way to go before we can expect AVs to be widely rolled out and part of a typical workday commute. But as software developers, manufacturers, city planners, and other stakeholders begin to think more seriously about how these vehicles could be introduced into society, it’s important to remember the cultural element.

For more on AI considerations in transportation and other sectors, check out some other APEX posts on the subject here and here.