If you think intelligent software such as Google Now is as cool as we do, hold your on to your hat because the future has a whole lot of intelligence in store. In fact, a lot of these systems have already been developed and, now, we just need to sit and wait for them to be deployed across various industries. The smart car is much more intelligent than you might imagine – smarter than K.I.T.T from Knight Rider. And it's going to talk to your smart house. And you. Your car will know your name, it will know your habits, and it will know more about itself than you do.
Yes, this is real, and we spoke with AT&T's assistant vice president of intelligent systems research from AT&T Labs-Research, Dr. Mazin Gilbert, to see what kind of exciting tech the future holds. No, we're not talking about 10, 20 years from now. We're talking about 2-3 years and technology that already works.
Q: Let's start with speech recognition and the connected car. What are you doing on those fronts?
A: The connected car is just one big opportunity. The connected car market is changing rapidly with 4G LTE in cars. A large percentage of cars will be connected in 2025.. 100 percent of them probably. So we start with having connectivity and we ask, "what can you do with it?" Really, the focus we've been looking at is what does this mean to the experience for consumers? While this is all happening, there are a lot of stats that show that distracted driving is a big issue.How do we revolutionize the car of the future?
The car will be a platform just like smartphones. How do we do that in a way that's safe and fun? And something that's going to control all these accidents and issues? We focus on intelligent systems in that area. Speech recognition is already in cars, but it's lousy because the processing power, memory and the user interface are limited. Cost is controlled in the car so it's hard to do something advanced. These systems now are trying to handle 10 words, 100 words, 500 words. Nobody talks like that. So by having the connectivity, the services we're working on, and with device and car maker partnerships, we're going to change that all.
Q: That sounds cool. But we already have phones, and super smart ones like the Moto X that let's us talk to them. So why do we care about in-car systems? Why not just use our phones all of the time?
A: There's going to be 3 modes for a connected experience, for different situations and contexts. You'll see different user preferences, cars and prices. One mode is what we have today today: I bring my device to the car and I use it. That's the problem we have with texting and driving. We can't stop people from doing that, we can't.
The second mode is the tethered mode. That just means that the car will provide you with an interface so you can plug in your smart device and now the display shows in your headboard. You don't have to control it, you can access it on a larger screen in the car headboard. That exists today. But again, there are huge issues with driver distraction.
The third mode is where the intelligence is really transparent and where it resides. Intelligence for a mobile device is in the network not the device. For example, Facebook isn't local, it's on the network. Voice commands use the network. You turn it off and it can't do anything. We want to make sure that the intelligence you carry with you is in the car, on your device, at home, wherever you are, it's always with you.
The device, whether it's the car, your phone, or your laptop at home, they're just access points. They all provide a different experience with different capabilities. We call that your connected life experience. You don't care where it resides, you have the device where you are wherever you go. It's the connection that's important.
Q: But Google Now knows when I'm in the car on the Moto X. It's great with speech recognition. How is your system smarter?
A: Google Now is an application that allows you to do voice to text. It's for the consumer. You can ask the weather and it works fine. But if you start using it for business apps and say, "What's the ticker symbol for Amazon, oh, and transfer $300 from this account to that account," Google Now isn't built for that. It's a general consumer service and it has been trained by a huge amount of data.
AT&T Watson, which powers our systems, isn't an application. It's a platform. It's the tech that helps you build an app like Google Now, but can also be used in healthcare, education, the connected car. And in the connected car, it will support vocabulary that you can't have from Google Now because our system learns and adapts. It can learn accents, for example. Not because it was built to, but because of the feedback from large amounts of data that comes from huge numbers of people with different dialects, accents, and more. Watson learns from you.
Q: Ok. So we have all of these devices. How's that going to make my life easier? If they're all on the network, my car is even on the network, what's that mean when I'm in the car? How smart are we talking here?
A: We're developing agents that are contextually aware, environmentally aware, and are also reactive and proactive, so they have predictive intelligence. These systems know when the person leaves their home, they sit in the car in their driver seat, and then the camera in the driver seat sees you driving the car, and, immediately, it personalizes the config in the car – your seat, radio, temperature, everything.
We know you left your home, so we check that the garage is shut, the door is shut, we do that by connecting to AT&T's Digital Life home automation services. The system told you it checked everything and then says good morning how are you today. You didn't say anything, in this case there's no need for speech recognition, the tech was all behind the curtain. This intelligent system did all that for you. Actively, without you knowing.
Now you're driving in the car and a sensor noticed that your light is broken. What happens today, it might light up an indicator telling you that a light is broken. What we're doing is not just that, it finds that the light is broken, it goes and checks whether your car dealer carries that light, checks your calendar to see if you're available to visit for a fix at a particular time, and it comes back and says your light is broken you can go to the dealer at 10:00 a.m. tomorrow. Then it asks if it should book the appointment for you.
It might know that you like Starbucks and you go there every morning. It knows that, so when there's something interesting when you're driving near a Starbucks, or walking, or whatever, if there's a discount the system will tell you. That's what we're trying to do. Build the intelligence into the network. It always follows you and is personalized to you. It's opt in, of course.
Q: Sounds awesome but pretty far out. I want it now. Can I get it now?
A: Connected cars are a brand new market, there are only like one or two models now. But the connected car market is expected to be a $60 billion to $80 billion market by by 2018. That includes manufacturing of the cars and everything else, platforms, software, hardware and headboards. It's new, that's why we're excited. The car will become a very smart platform. The intelligence required is completely hybrid. There will be some intelligence in the car, and some in the cloud. It doesn't matter where the intelligence resides, we don't care, everything has sensors.
AT&T is providing the LTE connectivity in these cars to help drive these experiences. That's a lot of what we've announced with many car makers. Without connectivity a car is stale. It's a box that doesn't talk to the outside. Today, cars take a minute and a half to input an address by voice. I could do that on mobile with AT&T TeleNav and get it done in about 3 seconds. The technology that's embedded in cars, that's already 2-5 years old.
The "future" is how consumers may view this connected technology, but it's here today. We're just working with car makers doing the integration. It takes 2-3 years for this stuff to roll out. But it's going to get more and more sophisticated, and even more rapid because of the connectivity. This isn't a vision of the future, it's something we're working on right now.
Q: It all sounds super expensive though. I mean car companies can nickle and dime us for these features. What if we could buy kits that turns our older cars into smart cars? Is that something that's possible?
A: The connectivity addresses the costs. It opens the gate. If you put all of this intelligence into a car without the cloud, it will cost you a lot of money. You'd need a lot of processors and memory and you wouldn't be able to afford the device. Put it on the network and it's affordable.
I assume third-party kits could happen. People don't change cars every two years like they do mobile devices. Once there's tech, it will move much faster than we've even seen with mobile. There's no way you're going to change your car every time something new comes out. Whether AT&T ends up providing [kits] or other vendors do, that's a question I can't answer. But that kind of idea is certainly expected.
Wrap-Up That concludes our chat with Mazin Gilbert. He also touched a bit more on Watson, which helps create a foundation for this platform – each part of it serves as sort of a jig-saw piece in the larger puzzle that's part of our connected future. And surprisingly, this isn't science fiction. Gilbert and his team work on this every day, and soon it will become our reality.