AI set to drive Uber's India plans with focus on local languages

Uber’s in-app calling feature through VoIP was hailed as a security measure that provided greater anonymity and privacy to both the rider and driver
Ride-hailing app Uber’s calls, routed through the Internet, will be used to train its internal artificial intelligence (AI) engines in local Indian languages, Business Standard has learnt. Uber began rolling out the in-app calling feature through voice over internet protocol (VoIP) in October last year. It was introduced in India early this year and was hailed as being a security measure that provided greater anonymity and privacy to both the rider and the driver.

Some drivers said they aren’t sure if conversations over the Internet made through the app are recorded. “It’s a good feature but we don’t have clarity on whether Uber records these calls. They might be using it for training purposes,” said an Uber driver who did not wish to be named.

The calls routed through Uber’s call centre and its app will also be used to train AI engines to improve local navigation in India, by identifying the various local hotspots and important landmarks, people in the know told Business Standard.

For regular users of Uber, snippets of recorded calls will be used to identify landmarks near their house or workplace, and other places they regularly visit. Apart from improving landmark identification, these recordings will also be used to improve the AI’s understanding of local and native languages. 

Once the AI engine starts understanding local languages, it will help the company increase its rider base. It does not want to stick to just English and Hindi, said one of the people. In its privacy policy available on its website, Uber states: “We enable users to communicate with each other and Uber through the Uber apps, websites, and other services. For example, we enable drivers and riders, and delivery partners and recipients, to call or text each other (in some countries, without disclosing their telephone numbers to each other). To provide this service, Uber receives some information regarding the calls or texts, including the date and time of the call/text, and the content of the communications. Uber may also use this information for customer support services (including to resolve disputes between users), for safety and security purposes, to improve our products and services and for analytics.”

The policy is from before the in-app calling began rolling out and does not have a specific update on these calls and whether and how they are recorded. From a user’s perspective, there is little option in the absence of a privacy or data protection law, and such collection and use of data fall under a grey zone.

“We have a fundamental right to privacy but that is enforceable only against the State. Private entities are non-State. In their capacity as intermediaries under the Information Technology Act, they have to exercise due diligence under Section 79C,” said Pavan Duggal, a cyber-law expert. He added that chances of misuse of data is high in such a scenario, and people need to be cautious while using such services.


There is also the issue of where this data is being stored. If it is on servers outside India, Indian laws will not be applicable.

In a bid to improve passenger safety, Uber India also introduced the phone number anonymity system recently, for the country under which the ride-sharing app routes calls to and from the driver (via its call centre)The capability was implemented on an urgent basis in India after complaints started coming in of driver partners storing numbers of female passengers and stalking or harassing them via calls and messages, people in the know who have worked on the project, told Business Standard.

The rider cannot share their number with the driver or vice versa, even if they opt for it, a spokesperson for the company said. When a rider and driver contact each other, both numbers are anonymised; you will not see each other’s personal contact details but can still call to discuss important details, like pick-up locations. This feature symbolises our continuous effort towards safety and ensuring a better experience for driver partners and riders, the spokesperson said in response to an emailed questionnaire.

The company did not respond to specific questions about storing voice data for training its AI system.