'We have the ability to build products that make people's lives better'

From his early days studying computer science and languages, Bertrand Damiba has always been excited about the intersections of technology and speech, human and computer-facilitated communication. And for nearly six years, he’s been pursuing his passion at Google as a product manager: “I believe in my heart that we build products that matter. I believe we have the ability to build products that make people’s lives better. That’s really motivating. It’s a belief that’s shared across most people that I work with here at Google.”

Damiba's team works on natural language generation for the Google Assistant. Natural language generation is an outgrowth of natural language processing. While there has been a lot of progress on getting machines to understand language, he explains, machines weren’t good at speaking in a way that mirrored what users do. In natural language generation, computers learn to find the best way to stitch together words to express answers in a conversational way.

For users to believe, to trust that they can speak naturally, we have to redefine their relationship with Google.

It’s not easy: “For us to be able to sell the metaphor of an assistant, it requires us to get a lot of things right. [People] need a lot of cues in order to believe that metaphor,” Bertrand explains, using the example of Serena Williams. To research Williams on a computer, you’d likely type in “Serena Williams.” To converse about her, you might say Serena Williams one minute and refer to “her” the next—”her” career, “her” engagement. You wouldn’t say Serena Williams each time the conversation progressed. It wouldn’t feel natural. The transition from “Serena Williams” to “her” is a figment of natural speech.

Put another way, he says, “I may use the same words you’re using when I reply to you. If you use the word ‘picture,’ I’m likely to use the word ‘picture’ too. We’re building this vocabulary within the conversation.” The conversational vocabulary is key to making users believe they can have a natural conversation with the Assistant. Bertrand points out that tech companies have spent years training people not to interact naturally with technology when they formulate search queries, for example. “For users to believe, to trust that they can speak naturally, we have to redefine their relationship with Google.”

The ‘Star Trek’ computer or the movie ‘Her’ have had an influence on what we aspire to deliver. We want something that’s as good as what we’ve seen in movies and tv shows.

In order to understand how people might experience the product, and how to create for a platform that doesn’t yet exist, the team tested it out among themselves and with other Googlers. The diverse backgrounds of the team members and testers helped improve the product. “French is my native language. A lot of us don’t have English as our native language. We have our own biases and tendencies. That’s sort of the richness the team brings to the equation,” says Bertrand, who was born in Burkina Faso. “Right now, on average, every person on my teams speaks two languages. We have an immigrant parent or we were born somewhere else, so we bring this to the fore when we test products. When we’re about to launch something, it sounds like the U.N.”

Imagining something completely new is also a perk of the job. “This is where I think the job becomes really interesting. A lot of my job is to bring in my personal experience as a human being. Obviously a lot is drawn from my colleagues,” he says. And then adds, “the ‘Star Trek’ computer or the movie ‘Her’ have had an influence on what we aspire to deliver. We want something that’s as good as what we’ve seen in movies and tv shows.”

As a product manager, Bertrand gathers data and collaborates to ensure an excellent user experience. “We work with designers, engineers, linguists. Everybody contributes. Everybody wants to build a great product,” he says.

That desire to build something great means feedback is taken seriously. For example, the Assistant supports questions like “How’s the market doing?” based on the feedback of early testers. Bertrand was also able to use the diverse backgrounds, interests, and experiences of team members to ensure the Assistant could converse with anyone on any topic, from the NASDAQ, to the NCAA, to hip hop.

Bertrand Damiba

Product Manager

'We have the ability to build products that make people's lives better'

From his early days studying computer science and languages, Bertrand Damiba has always been excited about the intersections of technology and speech, human and computer-facilitated communication. And for nearly six years, he’s been pursuing his passion at Google as a product manager: “I believe in my heart that we build products that matter. I believe we have the ability to build products that make people’s lives better. That’s really motivating. It’s a belief that’s shared across most people that I work with here at Google.”

Damiba's team works on natural language generation for the Google Assistant. Natural language generation is an outgrowth of natural language processing. While there has been a lot of progress on getting machines to understand language, he explains, machines weren’t good at speaking in a way that mirrored what users do. In natural language generation, computers learn to find the best way to stitch together words to express answers in a conversational way.

For users to believe, to trust that they can speak naturally, we have to redefine their relationship with Google.

It’s not easy: “For us to be able to sell the metaphor of an assistant, it requires us to get a lot of things right. [People] need a lot of cues in order to believe that metaphor,” Bertrand explains, using the example of Serena Williams. To research Williams on a computer, you’d likely type in “Serena Williams.” To converse about her, you might say Serena Williams one minute and refer to “her” the next—”her” career, “her” engagement. You wouldn’t say Serena Williams each time the conversation progressed. It wouldn’t feel natural. The transition from “Serena Williams” to “her” is a figment of natural speech.

Put another way, he says, “I may use the same words you’re using when I reply to you. If you use the word ‘picture,’ I’m likely to use the word ‘picture’ too. We’re building this vocabulary within the conversation.” The conversational vocabulary is key to making users believe they can have a natural conversation with the Assistant. Bertrand points out that tech companies have spent years training people not to interact naturally with technology when they formulate search queries, for example. “For users to believe, to trust that they can speak naturally, we have to redefine their relationship with Google.”

The ‘Star Trek’ computer or the movie ‘Her’ have had an influence on what we aspire to deliver. We want something that’s as good as what we’ve seen in movies and tv shows.

In order to understand how people might experience the product, and how to create for a platform that doesn’t yet exist, the team tested it out among themselves and with other Googlers. The diverse backgrounds of the team members and testers helped improve the product. “French is my native language. A lot of us don’t have English as our native language. We have our own biases and tendencies. That’s sort of the richness the team brings to the equation,” says Bertrand, who was born in Burkina Faso. “Right now, on average, every person on my teams speaks two languages. We have an immigrant parent or we were born somewhere else, so we bring this to the fore when we test products. When we’re about to launch something, it sounds like the U.N.”

Imagining something completely new is also a perk of the job. “This is where I think the job becomes really interesting. A lot of my job is to bring in my personal experience as a human being. Obviously a lot is drawn from my colleagues,” he says. And then adds, “the ‘Star Trek’ computer or the movie ‘Her’ have had an influence on what we aspire to deliver. We want something that’s as good as what we’ve seen in movies and tv shows.”

As a product manager, Bertrand gathers data and collaborates to ensure an excellent user experience. “We work with designers, engineers, linguists. Everybody contributes. Everybody wants to build a great product,” he says.

That desire to build something great means feedback is taken seriously. For example, the Assistant supports questions like “How’s the market doing?” based on the feedback of early testers. Bertrand was also able to use the diverse backgrounds, interests, and experiences of team members to ensure the Assistant could converse with anyone on any topic, from the NASDAQ, to the NCAA, to hip hop.