Just after New Year’s Day 2017, Appen embarked on a new venture: officially establishing an office in Detroit, Michigan. This new location has allowed us to better support our growing list of clients in the Motor City. Just as voice recognition technology has become a standard feature in new cars and trucks, there has been a parallel demand from customers for a better in-car voice experience.
Appen has risen to the challenge, lending our data and testing expertise to an increasing number of original equipment manufacturers (OEMs) and technology suppliers in the automotive industry. Last year, we recognized the opportunity to work more closely with those clients, and provide them with even better support, from a local office.
Heading up the effort is business development manager Adrian Buncuga. He started his career with Appen while finishing a degree in Linguistics at the University of Sydney, and has now been with us in various roles for over seven years. A New Zealander by birth, he’s lived in the US since 2013, and in Detroit since January.
So far, so good. Learning about Detroit’s rich cultural history has been Adrian’s favorite thing about the city so far, and he says he’s grateful for the neighborliness of its residents, who have helped make it a smooth transition to date. Next on the list, of course, is the local cuisine. “This city is made up of so many different cultural communities and they conveniently all seem to agree on a love of amazing food,” he says. “If I had to pick one thing, I’d say the garlic sauce at my favorite Lebanese spot is enough to keep me in the area for a long time.”
From the ground up
One of Adrian’s longer-term projects is to find, lease and build out the actual brick-and-mortar office. For the time being, he and two project managers are based at a shared workspace in Midtown Detroit called the Green Garage. “The full office build-out is on the way, but we’ve got plenty to keep us busy for now,” Adrian says. “We’re constantly meeting face-to-face with our clients and we’ve got a good mix of projects to manage, both remote and local. The most important thing is that we have the experience and resources available to quickly adapt our local operations to whatever our clients need.”
Current local projects are primarily focused on training and testing “head units,” which is industry-speak for the information and entertainment units in cars that you speak and listen to. Appen has also built a strong reputation for our high-quality, large-scale speech data collections. We’ve done those for 6 of the top 10 leading automotive OEMs, and we’ve just finished conducting some on-site recordings at one of our biggest clients’ facilities.
To provide a positive user experience for the maximum number of people, our clients’ speech interfaces must speak and understand many languages. Fortunately, Appen has worked in over 180 different languages, so we already have a great foundation of linguistic data and testing expertise to build on. As Detroit is home to so much diversity, Adrian and his team are also busy building out a Detroit-based crowd of native speakers of various languages, with localization and quality assurance (QA) experience, so that Appen can test and train these systems in as many languages as possible. He says, “These native speakers offer a ton of value for our local clients, who know that human-generated language data, combined with human testing, is still the best way to create systems that work.”
Increased safety, increased expectations
Why is the auto industry embracing voice recognition technology? “Certainly, safety is a big reason,” says Adrian. Now that the technology is available and more common, there’s increased trust in the systems and their efficacy among government agencies and consumer advocate organizations. Traditional dashboards—and more recently, mobile devices—take a driver’s hands and eyes off the road. Speech interfaces don’t.
Shifting consumer expectations are also playing a major role, with more people relying on artificial intelligence (AI) through the built-in assistant in their phones and at-home assistants like Amazon Echo and Google Home. “Consumers who invest in these kinds of products have increasingly high expectations about how they should be able to interact with their cars,” says Adrian.
The auto industry has opened the door to more AI-based improvements than just voice recognition, and Appen is gearing up to support them too. Autonomous driving is the most topical example right now. While self-driving cars are extremely complex machines, their AI is absolutely powered by machine learning. “As the car moves forward,” explains Adrian, “it must process a lot of visual data in front of it—just like a driver does when looking out the windshield. Appen can take large volumes of that image data, and assign meaning to what’s there, such as identifying a tree or pedestrian, and feed that back into the car’s AI to teach it. We can annotate this image data using an enormous crowd of trusted consultants. Our engineering team is also actively developing the platform that supports this work.”
“Sentiment analysis” is another interesting avenue for research and development (R&D) in the industry. It could prove to be an important next step, giving cars the ability to identify speakers’ emotions as well as their words—so they could tell when users get frustrated, and respond accordingly. “We’ve helped develop a lot of training data for sentiment analysis with clients in other tech sectors,” says Adrian, “and we’re very well poised to transfer our experience in that area to support our automotive clients as well.”
There are plenty of other ideas afoot. Appen is keeping an ear to the ground, talking with clients and ramping up our internal platforms and processes so we can handle whatever the future brings.
Six months into the official kickoff of Appen’s Detroit “office,” we’re all excited about the possibilities. “Detroit is definitely keeping me busy,” says Adrian. “I had 26 on-site visits in the first quarter. Right now, I’m excited to build on the great foundation Appen has with current clients, and develop new client relationships so we can learn more about what they’re working on and how we can help.”