Raised in a family of engineers in India, Rohit Prasad’s first research problem was finding a television on which to watch his favorite program: Star Trek. “This was at a time when TVs were still a rarity in India,” he said. “But Whenever” Star Trek I managed to find one and plant myself for it. Unlike many other Trekkies, Prasad was not a space nerd eager to explore new worlds and meet distant civilizations. He was fascinated by the show’s talking computer. “That great, all-knowing, voice-activated computer, which you could make requests to just by saying ‘Computer,'” he said. “It was very inspiring to me and made me want to build something like that – something that would change people’s lives.”
He did. Prasad has spent the past decade building Alexa, Amazon’s conversational AI, into a technology that now powers people in millions of homes around the world. Amazon recently announced that in addition to his role as chief scientist, Prasad is also the new leader of the Alexa business. “We’ve built a solid foundation for achieving our vision of Alexa as an indispensable personal assistant, advisor and companion for anyone, anywhere,” he said. “Alexa is one of the most loved and advanced AI applications in the world. I’m excited about Alexa’s adoption and excited to make it even more useful for customers in their everyday lives.”
We recently caught up with Prasad to learn more about his plans for Alexa and his passion for AI.
For many people, the talking computer is still science fiction. How do you make it come true?
I was inspired by the Star Trek computer and always wanted to help build one. So I focused on speech processing in graduate school and chose to work on conversational AI as my first job. For the past 10 years, our team has been developing Alexa, a service designed to be an indispensable trusted AI assistant, advisor and companion for everyone, everywhere. Today I feel like we are closer than ever to fulfilling that dream.
Where is Alexa now?
I still have to pinch myself. Alexa started as an idea on a whiteboard and somehow in less than a decade has evolved into an AI service that allows millions of customers to communicate billions of times a week in different languages and cultures around the world. Even in the past year, Alexa interactions have increased by more than 30%.
What is behind this growth?
Our work begins and ends for the benefit of our customers. We’ve doubled the experiences customers tell us they like the most. For example, when we first launched Alexa, customers could use the service to do just a handful of things, like play music and set timers and alarms. Today, you can use it to control your smart home, watch videos and shows, shop, learn about any subject and much more with over 130,000 skills available to our customers. All of this is made possible by Alexa’s AI, one of the most complex AI systems with over 30 different machine learning systems. A major reason for this growth is not only how much better Alexa has become at responding to customer requests, but also how Alexa initiates actions on behalf of customers, which we call proactive assistance.
Can you give us some examples of proactive help?
Alexa Hunches and Routines are two great examples of how Alexa has revolutionized everyday convenience. They enable Alexa to proactively anticipate your needs and automate your day. I love my Alexa morning routine: when I turn off my alarm, Alexa automatically turns on my bedside lamp, tells me the day’s weather, and launches NPR Morning Edition. And with Hunches, I have peace of mind that if I leave the garage door open, say, Alexa can remind me or close the door itself. It’s pretty magical. More than 30% of Alexa actions in smart home control are now initiated by Alexa.
Another example is in the entertainment area. We recently launched a conversation experience called ‘What should I watch?’ It combines Alexa and Fire TV recommendations to make Alexa an entertainment expert that can help you find content you might like on your Echo Show and Fire TV. Thursday night soccer in the US is another example. We’ve launched features so you can ask Alexa for the latest in-game player and team stats without interrupting the game’s live stream.
What about shopping with Alexa?
Shopping is another great place to see the evolution of Alexa. In the beginning you could add items to your shopping list. With just your voice, Alexa can help you place orders, get product questions answered, and get delivery notifications. We also recently announced a new feature – coming soon – called Shop the Look. It uses an AI breakthrough in multimodal understanding, allowing you to browse fashion, home and beauty products directly on your Echo Show by saying things like, “Alexa, show me some green yoga pants” to display products on the screen. to choose. People love shopping with Alexa. Fifty percent of Alexa customers have used their device for shopping.
Can you tell us about your future plans for Alexa and give us a glimpse into what your team is working on?
Alexa is getting smarter every day, but we’re still a long way from the final frontier. Today, you can use Alexa to do things ranging from entertainment and requesting information to controlling and automating your smart home, shopping and much more. But Alexa is already much more than your voice assistant. It is also an expert advisor and for some even a companion. When it informs you that you left your thermostat on 72 degrees when no one was home, it proactively advises you to save energy. Certain customers, especially children and older customers, have a companion-like relationship with Alexa and Kids+ on Alexa and have much longer conversations than single turn requests.
The way to be great at meeting the ever-growing expectations of Alexa is through what we call generalized intelligence, which is when AI is able to learn multiple tasks and constantly adapt with limited human supervision. Alexa is not only one of the most fertile testing grounds for advancing generalized intelligence, but also for developing human-like interaction skills beyond speech interactions.
What do you mean?
Communication between people, and hopefully soon with Alexa, is so much more than just voice – it’s facial expressions, hand gestures, body language. That’s why we’re investing not only in improving Alexa’s voice AI, but also in multimodal AI, improving Alexa’s ability to process various sensory signals such as images, touch and ultrasound, as well as speech. More natural interaction makes Alexa an even better assistant, advisor and companion. This is the hallmark of a conversational AI application like Alexa.
In addition to conversational AI, you often refer to ambient intelligence. How does ambient intelligence apply to Alexa?
Ambient intelligence is the science that allows the various devices, sensors, and technologies in your environment to work together seamlessly to assist and delight you at all times. This intelligence is there when you need it—it even anticipates your needs—but it fades into the background when it’s not needed. With features like Routines and Hunches that help Alexa anticipate your needs and automate your day, Alexa is at its core what Amazon does with ambient intelligence. (View and read our environmental intelligence explanation.)
What is your vision for the future?
I’m sure it’s counterintuitive to hear from someone who has built their career in consumer technology and AI, but I think we should think less about technology. Instead, we should focus on experiences that make your life simpler, not more complex. We want to free up hours in our customers’ everyday lives and reduce their cognitive load so they can spend more time on the people and things that matter most to them.
In many ways, today’s Alexa is already much more powerful than that Star Trek computer that inspired me. But, as we say at Amazon, it’s still day 1. We have an ambitious vision and I’m optimistic that we will continue to delight our customers and deliver more value.
