The New Google Assistant Is Scary Smart

We have an exclusive here from 9to5Google. Here's how the new Google Assistant on the Pixel 4 is going to work. And so there's actually some leaked videos here of the brand new Google Assistant. Of course, there's been all kinds of leaks around the Pixel 4 hardware. Now we have some leaks around the software as well. If you scroll down, Will, you'll see there's a couple of video clips. The first one, two girls, they're kind of running down the street. They're late, I guess. She squeezes, I think, to launch Google Assistant. And she says, text Dan, we're late. Okay, so Google kind of showcasing how it's going to integrate with different applications. Hey, Google, turn on do not disturb. The guy, he's looking, he's at the lake, he's at the cottage, and so forth. Okay, so that's not too crazy. It's like, all right, I understand what's going on there. But you scroll down a little further and you start to see some of the real heavy lifting going on. Here's a text from someone that says, what time does your flight land? Right within Assistant, the guy, I assume it's a guy, a guy or girl, checks the flight time. You see, they check the flight time because, of course, Google Assistant knows everything. And then Google Assistant replies with the relevant information right in the thread. Reply, I get in at 514. Pause that for a second. The smart part here is the fact that Google Assistant knew the context. It knew that this recent text that came in with the word flight in it would be the place you would want to respond with your flight details, such as when you get in, as opposed to needing to say, text Tori, and that's her name's Tori in the demo here. And then having to, you're starting the process over and over again in the, this is called continued conversation, where Assistant is smart enough to understand a certain amount of context as to what you mean, and also to continue tasks that you started as opposed to having to end and start new ones. You understand? Now, this feature was brought over from Google Home and Smart Display, and it allows you to have back-to-back commands in context of one another. Now, the next demo here, it showcases Assistant's ability to control specific applications on photos. So show me, so if you actually go to the very front of this clip, Will, the very front of it. Okay. So the command here is show me photos from last weekend. Now you can hit play. Show me my photos from last weekend. Boom. Brings up your photos from last weekend. Show me the ones with a dog. Okay, that's important. It didn't say, show me photos with a dog. Now it's conversational. It knows your info. It knows your previous command. So it's using that to determine what your next command is probably saying. Show me the ones with a dog. Keep playing. Boom. All the photos that featured a dog, the AI, it's sifting through. It's looking for dogs. And then boom, straight into the text. So tons of context here. Apparently, a lot of this was getting loaded onto the phone. They were capable of shrinking voice recognition and understanding models that at one time were around 100 gigs in size. So they couldn't live locally on the phone in most cases without taking up too much space. They shrunk all that down into 0.5 gigs. And by moving it directly onto the phone, they were able to increase the speed, lower the latency so that you can have this conversational thing without feeling the lag. So it's not in the cloud. Not in the cloud. Not in this particular case. Not for this set of commands. It'll allow you to interact with such things like turning on and off airplane mode, flashlight, all kinds of complex system-based actions. Of course, Google releasing its own hardware partially helps in enabling this without needing permission for all this stuff. So it'll be really interesting to see how these integrations map out and just how complex you can get with your set of instructions. I see this in the future being a major differentiating factor, how the assistants operate as we continue to interact more and more through voice and through our own personal AIs, essentially understanding context and knowing what we want to do before we want to do it. I mean, it's pretty amazing potential future there. And this is one of the areas in which you have to believe companies like Apple, Amazon, and others are looking at this saying, okay, we need to get up to this particular standard because it is, you can imagine, what's the movie that you always tell me to watch? I haven't seen it yet. Her? Her, yeah. You can imagine there's the potential for far more audio and voice interaction between us and our smart devices than we're currently taking advantage of. It's almost like a conversation where everything is seamless. And you know that's where it's going. Yeah. But you can imagine a circumstance in which you get into your car, turn on this song, turn off that song, call this person. The person doesn't pick up. Do you want to send them a message instead? Send them this message. I'd like to travel to this particular place. Oh, there's an accident in that particular place. Do you want to take this other route? That other route? Just this ongoing conversation that you could have with your AI. A lot of those things I just mentioned already exist. But you imagine a situation in which it's normal to converse with your personal AI. And now things are starting to get in that science fiction zone that I'm always talking about. It could be terrifying. It could be incredible. It could free you up more. I think that's why they showcased in this demo the moment of do not disturb. Imagine if your AI knew that you're at the cottage, looks at your location data. It's like he's looking at a lake right now. Imagine it was like, do you want me to turn on do not disturb? And you didn't even think about it. You were about to be disrupted. And you're like, yeah, do that. And boom, now silence. You enjoy the moment. It could be helpful too. It doesn't always need to be dark and grim. It could actually help you engage with your smartphone less. No, for real, Will. I know you're looking at me. You're like, I don't know about that, Lou, but it could. Because right now it's multiple steps to do everything. And once you're looking at the screen, you know you're deep. You're hooked. Yeah. I think if we could do more things with voice, we actually might manage our digital situation a little bit better. Who knows? I'm just saying. I'm putting it out there as a potential. Exactly like the movie. You got to check it out. Okay, I'll check out the movie. I'll check out the movie.