Connect with us


Baby You Can Drive My Car: How Cars and AI May Predict Health

 4 min read / 

Sedan = Democrat. Pick-up = Republican. The method used to arrive at this conclusion is really more fascinating than the answer and provides, perhaps, a view into the ways Big Data will in the future be used to understand all types of consumer preferences to inform marketing decisions and how the information trail we all leave behind online and in our driveways enlightens artificial intelligence (AI).

Researchers recently reported how they taught and then used AI to scrutinise 50 million Google Street View images to find 22 million cars to build more than 2,600 automobile categories to find out if you voted Democrat or Republican. The idea was to learn if socioeconomic status and political persuasion could be gleaned from identifying what was parked in your driveway.

Could they do it? Yes and quickly.

Two weeks after putting the question to the AI, researchers could accurately predict if a community would vote Democrat or Republican just by looking at parked cars. Communities with a preponderance of sedans had an 88% chance of voting Democrat; neighbourhoods with extended cab pick-ups had an 82% chance of voting Republican.

“(T)he vehicular feature that was most strongly associated with Democratic precincts was sedans, whereas Republican precincts were most strongly associated with extended-cab pickup trucks (a truck with rear-seat access). We found that by driving through a city while counting sedans and pickup trucks, it is possible to reliably determine whether the city voted Democratic or Republican….”

They never knocked on any doors. They did not go through voter records. They did not use a phone to conduct surveys. The AI simply looked at a picture of an automobile. “We show that socioeconomic attributes such as income, race, education, and voting patterns can be inferred from cars,” they wrote.

Your Car, Your Health

Combining a car/socioeconomic status may be a predictor of overall health and wellness. Socioeconomic status has long been associated with overall health. In general, the lower the socioeconomic status, the worse off one is health-wise. This is true for adults and teens.

Returning to the car research, it could be possible to delve into the 2,600 automobile categories and look closely and make and model, which could potentially elicit the socioeconomic status of the owners of each car. Imagine researchers or companies making similar inferences, but about your health, using AI, auto photos and your online footprint.

Would it be possible, then, to attach a health status to individuals, families or communities without ever talking to them? And what would we do with that information?

Big Data, Big Opportunity

Data mining companies already collect information about people, which is then used to understand how to serve specific online ads to specific users. Every piece of a person’s online presence can be examined. Whether they are conversing with friends on Facebook about health issues or shopping on Amazon for over-the-counter remedies, every digital step is recorded, categorised and analysed, along with the user’s IP address, which relates to a physical location.

So a computer could be trained to digest online data and offline data and then regurgitate a well-thought-out and, very likely, highly accurate socioeconomic status for a neighbourhood. The information about automobile choices and how much they cost could inform, for instance, pharmaceutical companies, local gyms, elective surgery centres and other health-related enterprises on how much money to spend in certain areas. The ads could be served up on a smartphone, tablet or any Internet of Things connected device sitting on kitchen counters worldwide.

Car Alarm

Admittedly, it is somewhat alarming to know so much can be gleaned from just a quick glance at cars and trips online. No matter how this scenario plays out in the future, caution is key. Health information security and general privacy must be maintained, even as companies use the information to provide better services and improved experiences. An unintended consequence, however, could be attributing specific health issues and conditions to neighbourhoods that are wholly incorrect.

The automobile researchers provide a very cogent warning:

“Although automated methods could be powerful resources for both researchers and policymakers, their progress will raise important ethical concerns; it is clear that public data should not be used to compromise reasonable privacy expectations of individual citizens, and this will be a central concern moving forward.”

Sign up to Mogul News.

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Send this to a friend