A decade ago, nobody was very familiar with mobile-first companies, let alone cab-ordering services. But 2011 not only brought with it mobile-first companies but also ride-hailing services. When you combine both of those, there is one company that comes to mind, Uber. Uber is one of such companies that has made quite a name for itself. This app has made all our lives easier since one can order a cab for themselves whenever they need it, whether it is an emergency or to go to work. Mobile-first companies such as Uber have built their business around mobiles.
Uber is one of the ride-hailing companies that operate worldwide. Though Uber launched in 2009 in the United States, it didn’t start running in other countries immediately. The app and its services first went international in 2011 when they first extended their services to Paris. Over time, Uber has brought in many developments and has also started using AI for better ride experiences for their customers. Uber does have many rivals, such as Lyft, which is also based in the US. Still, it is the first-ever taxi company which has started improving their services with artificial intelligence.
Contents
AI features Uber uses for each ride
At present, Uber does not just focus on mobile phones as being their core concept. They have now started to focus on artificial intelligence as well. It is a core concept, and there is no ambiguity whether it should be used or not. Jairam Ranganathan, the product director at Uber, stated that they don’t even think about it anymore. It goes without question that they must use AI. He said it is much like “not thinking about computers.”
Simply put, machine learning, artificial intelligence, neural networks, etc., are now the most crucial parts of every business process at Uber. Ranganathan mused in an AI-focused conference that they used AI in all aspects of its application. Their most significant success has been in the customer support system.
Uber uses artificial intelligence for applications, including safety processes, fraud detection, matching riders to drivers, risk assessment, driver onboarding, route optimization, etc. Some of the features where Uber uses AI are listed below.
Communication with the driver by One-Click Chat
Suppose you are standing on a curb waiting for your cab. Even as time passes, you see that the car icon is not swaying one bit from a particular location. That would be incredibly frustrating to you as a rider. You would naturally want to text the driver asking if everything is okay. But if the Uber driver has you text you back while he’s driving, it would be a safety risk to him.
Moreover, texting and driving are illegal in many jurisdictions. That is why the data scientists at Uber developed “One-Click Chat.” One-Click Chat is a feature on UberChat that enables both riders and drivers to send personalized responses that are fast and dynamic. This feature comes from a combination of supervised, that is, machine-learning and unsupervised methods.
How does it work?
When a rider sends a message to a driver, the back-end service by Uber immediately sends the message to the machine learning platform of Uber called Michaelangelo. This platform then uses NLP or Natural Language Processing methods. NLP is a method by which the medium of Michaelangelo pre-processes and encrypts the rider’s message and then evaluates its intent by using prediction scores. The service then provides only the top four replies that the platform suggested based on the prediction scores.
The replies are indicated by the platform Michaelangelo using a policy of reply retrieval. Finally, this message is sent to the driver, who can respond to the rider with one click. Intent detection is thus the crucial component when it comes to the working of “One-Click Chat.” To test this feature, Uber scientists had to first train a Doc2vec model on several millions of UberChat anonymized messages.
Figuring out each price hike
Uber admitted that it uses AI to charge its riders according to the price they would be willing to pay for each ride. According to Bloomberg’s report in 2017, Uber said that their newest system revolved around the concept of AI. It said that Uber employs AI and such algorithms, which estimated fare rates that several riders would pay without resistance based on the destination and time of day. In the past, the system generated uber fares depending on the time, mileage, and destination’s geographical demand. Whereas the new system is based on “route-based pricing.” It employs a machine-learning method to cause a difference between the pricing depending on various sociological factors.
A discrepancy of the new system
The new pricing system of Uber initially operated in 14 cities of the US but has expanded to other places, but there has been a particular discrepancy in the system. Since there was a tweak in pricing after Uber first implemented the AI system, there had been a price hike. But most of the time, the drivers don’t get the exact amount that a rider pays. Drivers have claimed quite often that they are yet to receive additional compensation from the hiked prices brought in by the AI system. Seemingly, Uber pocketed the excess amount caused by the gap between what a rider pays and what Uber pays a driver. To ease the drivers’ concerns, Uber decided to be completely transparent on what a rider pays. They also updated their service agreement and laid down the newest system’s whole structure, including AI.
Conclusion
Uber has brought in a lot of newer developments since they first employed AI in their systems. They now use AI to improve the accuracy of locations with perception and sense. The data team at Uber overcame the limitations that GPS had and ensured that locations were as precise as possible. They also incorporated computer vision into their process so that riding with Uber proved to be efficient and safe. Uber now plans to collaborate with more sources and services with AI. They are looking towards the new decade with plans to innovate and improve AI applications throughout their processes.
