Google has shared how it’s using artificial intelligence (AI) to try and keep up-to-date business hours updates on Google Maps, including its restaurant-calling Duplex technology, The Verge reports. The company said it would update the information on the map if it had enough confidence in the AI’s predictions of business hours.
In a blog post, Google outlined various factors that its artificial intelligence analyzes to determine whether it should make these updates. First, it looks at the time the business profile was last updated, the time of other similar stores, and Popular Times data to determine the likelihood that the time is incorrect. For example: if Google sees a lot of people visiting the store when it is supposedly closed, that could be a warning.
In its blog post, Google said its artificial intelligence looks at more data if it determines the time should be updated. It will take into account information from a business website, and even grab a Street View image (maybe a sign showing hours of operation) to try and figure out when a business is open. Google says it also checks with humans, including Google Maps users and business owners, to validate the AI’s predictions — and the company says it’s even using Duplex in some countries to ask businesses directly for hours of operation.
Google spokesman Genevieve Park told The Verge that Google will “only release when we have a high degree of confidence in the accuracy of our business hours.” If the AI thinks the time may be incorrect, but doesn’t have a reliable prediction, it adds a notification that the time may have changed.
Park also said that Google did not explicitly tell users that the time was updated by its artificial intelligence, explaining that artificial intelligence is used almost everywhere else in Google Maps. It looks like Google is quite bullish on its AI-driven approach. In its post, the company said it is “on track to update business hours for more than 20 million businesses worldwide over the next six months.”
Google also said it was piloting the use of another type of artificial intelligence in maps to help keep speed limits up to date. In the U.S., it will try to see if its partners have taken images of sections of road with speed limit signs, and will let artificial intelligence help its operations teams identify signs and speed limits posted on them.
While it’s not surprising that Google uses artificial intelligence to solve these problems, it’s also interesting to see so many interlocking systems involved. There’s computer vision, pattern recognition for location trends, and analyzing data from similar locations (which of course involves figuring out what similar locations are even), all in an effort to quietly try and keep up with how often businesses change time and Make sure it knows the speed limit for certain sections of the road.