The rise of platforms offering to connect buyers and sellers in everything from home maintenance to transportation has been one of the most profound economic changes in recent times.  These platforms are often commented on in terms of their tremendous scale and global reach, with articles often breathlessly exclaiming that Airbnb has more rooms than even the biggest hotel chains in the world.

A recent paper from Purdue University explored just how valid that assertion is.  The researchers examined a decade of reviews left for Airbnb and Yelp alongside employment data from the US Bureau of Labor Statistics to try and ascertain just what impact Airbnb was having on the restaurant scene in New York.

The data revealed that Airbnb guests staying in majority black neighborhoods seemed to be less likely to frequent local restaurants than guests staying in predominantly white neighborhoods.  What’s more, they were also around half as likely to return to the area a second time.  The reviews such travelers left would often describe the areas as dangerous or risky, despite these perceptions often showing a disconnect with official crime statistics.

“The review contents reveal that Airbnb visitors are more likely to discuss negative aspects of a local area that relate to safety if they are staying in a predominately black zipcode,” the authors explain. “Airbnb visitors may come to a black area without fully understanding the demographic makeup of a specific location. Over 60% of Airbnb hosts in predominately black areas, for example, are not black themselves.”

Societal biases

The data found that generally neigborhoods would enjoy a boost in restaurant employment whenever there was a boost in Airbnb activity, with the authors suggesting this could contribute as much as $1 million in extra tourism revenue.  This boost was generally absent however when over half of the residents in the neighborhood were either black or Hispanic.

What’s more, seemingly complementary technologies can only exacerbate this problem, as when people use an Uber to get around or search online for restaurants to eat in, they often fail to really engage with the local community at all.

“Technology allows us to find a lot more nuanced information about areas before we go there, and when you take a collection of technologies, then it becomes even more challenging,” the authors explain. “We’re lowering the cost of error with all these technologies, but in the process, we’re also losing the serendipity of unknown, of giving someone a chance. That’s out, that’s gone or that’s minimized today.”

Imperfect technology

The study highlights the imperfect nature of modern technology, and indeed the risk that technologies merely mimic the biases and prejudices humans suffer from.

As digital assistants and other technologies have an ever greater role in the decisions and purchases we make, it’s vital that the data these technologies are fed are as free from bias as possible.  If they’re not, then it’s only likely to results in high levels of bias in our consumption patterns.