ARI SHAPIRO, HOST:
Now it's time for All Tech Considered.
(SOUNDBITE OF MUSIC)
SHAPIRO: We've spent the last few weeks taking stock of the changes 2018 brought to the tech world and our relationship with the technology we use every day.
MARY LOUISE KELLY, HOST:
Now let's look at one particular advancement. The voice of artificial intelligence took a big step forward this year.
SIRI: My name is Siri.
CORTANA: Cortana here. How can I help?
ALEXA: My name is Alexa.
KELLY: Those voices - Apple's Siri, Microsoft's Cortana, Amazon's Alexa - they're all familiar to many of us by now.
SHAPIRO: And let's not forget Google's Assistant. It now uses a technology called Duplex, not only to recite information, but to perform tasks for you like call a hair salon or make a restaurant reservation. The website VentureBeat tried it out.
(SOUNDBITE OF ARCHIVED RECORDING)
COMPUTER-GENERATED VOICE: Hi. I'm calling to make a reservation for a client. I'm calling from Google, so the call may be recorded. Can I book a table for tomorrow, please?
UNIDENTIFIED PERSON: Yes. At what time would you like to reserve?
COMPUTER-GENERATED VOICE: At 7:00 p.m.
UNIDENTIFIED PERSON: How many people?
COMPUTER-GENERATED VOICE: It's for two people.
KELLY: That's Google Duplex. Google unveiled the technology in the spring, and this month, rolled it out for some of us humans to try. To talk about why this development is significant and where it might take us in the future, we are joined by tech writer Annalee Newitz. Annalee, welcome.
ANNALEE NEWITZ: Thanks for having me.
KELLY: You've called this the biggest thing in tech all year - how come? Why is it such a big deal?
NEWITZ: This is really something that builds on Siri and Alexa and other kinds of digital assistants. So with those kinds of assistants, you're talking with a machine. This is a way to talk to a machine and then have that machine talk to other people. So it's basically the next stage in having computers be conversational.
This makes computers much more accessible for people who may not be technical, who aren't able to use a keyboard for whatever reason. And so I think that this is less of a leap forward in AI, which is sort of how a lot of people have looked at it, and more a leap forward in making computers more usable.
KELLY: And how many people are actually using Duplex at this point?
NEWITZ: So it's not called Duplex. You can't kind of call up your little assistant in the phone. You actually have to be trying to make a restaurant reservation. And the idea is that if a restaurant won't accept online reservations, the assistant will seamlessly - behind the scenes, you won't even know it's doing it - it will call the restaurant for you and make reservations by phone. It's all happening behind the scenes.
KELLY: It the future that this could negotiate on your behalf? Say, right now, it calls the restaurant, asks for a table at 7:30, and the restaurant says, we got nothing. Then it could say, well, how about 8? Well, how about that little one by the back door? Well, how about - that it could reason on your behalf.
NEWITZ: Yes. Eventually, the idea would be that these kinds of assistants could be a representative for you. You know, if we look far into the future, far beyond 2019, what these kinds of devices might do is maybe attend business meetings for you...
KELLY: Oh, that would be good.
NEWITZ: ...Or go to a work meeting. That's when I think we start getting into really interesting legal issues because if your assistant isn't just haggling over a time for a table, but is instead haggling over a price or over a legal contract, that's an interesting question about liability because if your AI assistant does something on your behalf and it turns out to be illegal, whose fault is that?
KELLY: Fascinating.
NEWITZ: There's still a lot to work out there. I think that people do need to be informed when they're talking to a person versus talking to a machine.
KELLY: And why, why is that important in 2018, 2019?
NEWITZ: I think partly it's just psychological. We want to know who we're talking to. And we want to understand that this isn't a person. But I also think that, as time goes on, this is going to be a question of liability because what if my phone makes a mistake and makes a reservation for me and then it's one of those restaurants that says, if you don't make your reservation, you still have to pay 50 bucks? So now I'm on the hook for 50 bucks.
If we know we're talking to a person, we assume that they are taking some responsibility for their actions. If we know we're talking to a robot, we might want to double check on that reservation to make sure it wasn't accidental, it wasn't like a reservation butt call.
KELLY: (Laughter) I mean, I have to say, I'm exhausted a little bit just imagining the prospects of having to untangle some restaurant reservation that my robot phone butt dialed. Do you think, in the end, this will actually make our lives easier and will save us all time?
NEWITZ: I think that this will make some aspects of our lives much easier, probably not in making restaurant reservations. I think the best way that this will make computers more accessible for people is for people who have disabilities where they need to be able to talk to a machine instead of using a keyboard. Will it make it easier for us to order things online? I don't think so. I think it's already pretty easy.
So I think the real breakthrough here is that user friendliness, the way that it makes computers feel easier to use. But I am anxious about those questions around liability. When these assistants start making business deals for me, then I think I'm going to be a little concerned.
KELLY: Thanks so much, Annalee.
NEWITZ: Thank you.
KELLY: Annalee Newitz is a technology journalist, also an editor-at-large for the website Ars Technica. Transcript provided by NPR, Copyright NPR.