Motley Fool Money - Amazon’s Mistake in Voice Technology

Episode Date: November 12, 2023

You may be asking the wrong question about the future – it’s not what it looks like, but what it sounds like.  Tobias Dengel is the President of WillowTree, a TELUS International Company, that specializes in digital product design and development. He’s also the author of “The Sound of the Future: The Coming Age of Voice Technology.” Ricky Mulvey caught up with Dengel to discuss: What Alexa and Siri get wrong about voice interaction, Secret leaders in voice technology, And the next battle in the cybersecurity arms race. Tickers discussed: MCD, SG, AMZN, APPL, GOOG, MSFT, META, NVDA, TU Host: Ricky Mulvey Guest: Tobias Dengel Producer: Mary Long Engineer: Rick Engdahl Learn more about your ad choices. Visit megaphone.fm/adchoices

Hi everyone, I'm Charlie Cox. Join us on Disney Plus as we talk with the cast and crew of Marvel Television's Daredevil Born Again. What haven't you gotten to do as Daredevil? Being the Avengers. Charlie and Vincent came to play. I get emotional when I think about it. One of the great finale of any episode we've ever done. We are going to play Truth or Daredevil.

What? Oh boy. Fantastic. You guys go hard, man. Daredevil Born Again, official podcast Tuesdays, and stream Season 2 of Marvel Television's Daredevil Born Again on Disney Plus. The same thing's going on with voice. We've been messing around with it for 10 years.

We all want to use it because it's so fast to speak. We can speak three times faster than we can type, so it's just so efficient. But we don't want to listen. And the core mistake that all these systems have made, starting with Alexa and Siri, is that they try to create assistant human-to-human voice environments. They try to replicate those. But that's how what we want. We just want to say something and have the machine do something or give us the information.

and that's when voice will have the real interface breakthrough that it hasn't quite had yet. We all have it, but we don't use it for very much yet. I'm Mary Long, and that's Tobias Dengel. He's the president of Willow Tree, a global leader in digital design and development that's part of Telas International. He's also the author of the book The Sound of the Future. Ricky Mulvey caught up with Dengal to talk about voice technology and its potential to transform just about every industry, from food to banking.

They also discuss the original and future visions for Siri, the secret leader in voice, and what a Taylor Swift remix reveals about a cybersecurity arms race. I got a list of questions, but you said there's a connection basically to the start of the Motley Fool and the start of voice technology. That seems like a good enough point to jump off for this conversation. What are you talking about with that? Yeah, so one of my first meetings ever when I started AOL in 1996 was with David and Tom Gardner when they were just starting the Motley Fool. And what's interesting about that is at that

time, AWL was kind of the center of the Internet. So we would get to meet companies like the Motley Fool and TripAdvisor that would go on to change their industry, right? And there were a lot of, you know, incumbents who could have done that, but they didn't. It's always the new guys that disrupted. And I always compare that to a meeting I had about the same time with the executives of Coca-Cola who told us, I don't know how the Internet is ever going to impact what Coca-Cola. Cola does, right? And so, when these big technology shifts happen, like is happening with Voice right now, it opens up the world to a lot of brand new entrants and a lot of innovation. Yeah, so I know you help now, essentially what is it, helping disrupt the disruptors.

Is this a similar story playing out with some of the old guard as they welcome new voice technology where it's hesitant, they're hesitant, they don't really see how it'll impact anything? Or are you seeing a greater acceptance among those established brands? like your Coca-Cola's, like your General Motors, pick your old established. Yeah, I think in theory, there's a lot more acceptance right now because so many of these old brands got burned by the internet and then got burned by mobile. And so they're always looking for the next thing. Now, it's always hard to figure out what exactly is the next thing because 10 years ago,

people thought it was 3D TV and then people thought it was AR, Google Glass, and those things haven't happened. So you always got to pick where you really invest. But in theory, all these large companies want to invest in voice, it's more of a question of how rather than if. So then how are you seeing that breakdown? Because in theory is a critical word there. So where's the breakdown between in theory and in practice from your perspective?

Most of these voice technologies, if implemented well, are fairly disruptive to the existing status quo. So take a McDonald's, which is well publicized, been making a lot of voice investments. it's taken to its largest extent, it is going to fundamentally change how a McDonald's restaurant operates, right? Like the whole concept of walking in, talking to someone or talking to someone in drive-thru can change. So if you can do everything voice-based in an app before you even show up there, the whole

building might have to get redesigned, et cetera, et cetera. And so those are huge investments and huge disruptions. And usually they happen from the outside in an industry because there's so much change involved. Yeah, so now that we're on restaurants, I want to get to the top dogs and I want to talk to about the big players. But I think restaurants, there's some key disagreements here because a lot of automation has been promised for the restaurant industry and it's been very difficult to implement. We've seen things like the kitchen for Sweet Green where it's like this robot that makes

salads and it mixes it up together and there's very few people in the kitchen. But there's also been promises like, what is it, Flippy, the Robot. I don't know if you remember that one. is going to flip hamburgers and make it for you and you don't need anyone on a griddle again. We actually had Ron Shake on the show. He's the former CEO of Panera Bread. And he said the following about AI and essentially innovation in the restaurant industry. Quote, the food industry is the second oldest profession in the world. It's been around a long time. People are coming for an experience. People are coming in to socialize, to connect.

Ultimately, I heard about the demise. I heard in the pandemic, everybody was going to do delivery. that's going to be the way of the future. The cafe is going to go away. Dining experiences are going to go away. Now I hear about AI and I hear about technology. Nobody ever went to a restaurant for its technology. They go for the food and they go for the experience, end quote. I'll stop there and let you respond to it because it sounds like you have a different perspective. Well, I don't think so. I think it's just a difference of how technology gets implemented, right? I think it has to be in a human first kind of way. And there's also lots of segments within the restaurant. industry. Like if you take the QSR, quick serve, right, it's very, very price conscious. It is not about experience. You don't go to McDonald's necessarily for the experience. You go because of the price point and the quality relative to price. And voice is going to be part of making that technology, making those experiences even more efficient from a price perspective. So I think in QSR, I do have a different point of view. When you go to more of a casual environment, I think he's 100% right. That doesn't mean that technology won't play a role. There's no

reason that I have to tell my waiter or waitress what I want to eat, and then they spend the next three minutes trying to enter that into some either keyboard or whatever, or device that they handheld device they have with them. As soon as I say it, that information is out in the wild and should be immediately transcribed and used by the system. And I think that's a delightful experience for everybody. I'm going to go back to the quick serve, the fast food scenario. You're going to a drive-thru, and I think this makes a lot of sense. You're essentially, you're not just like talking to a bot. You're inputting the information you want, the order you want from the restaurant, the burger, the fries, the size of the drink you'd like, and then that shows up on a screen.

So you're not necessarily talking to like an AI bot necessarily, but it's the melding of the multimodal experience to get things a little bit more accurate and visual. Yeah, I think one of our key insights, and we've been studying this for a while, we went in 2015, 2016, we're looking at this mass adoption of Alexa and the previous adoption of Syria. We're saying, why do people want to do this? But at the same time, it isn't changing the world the way mobile did or the internet did. And the reason is the design was all wrong.

It was all these bot designs and humans don't want that. What we want is to give instructions and in the food example or a movie example, real time, see the order being created as we talk to an app or another screen. And then at the end of it, we just say confirmed and we're ready to go. What's amazing there is we're taking conversation, which is always call and response for millennia. And we have concurrent communication now where we're able to transmit information at the exact same time that we're able to receive it.

And that's just going to be a massive breakthrough. So we'll move now from the quick serve to the more, let's say the slightly less casual or the more sit-down experience, because I think this is where the disagreement may come into play. The current system, and this has been going around, I would say, for probably over a thousand years, right? So someone comes up to your table, they take your order, they deliver that back to the bar, to the kitchen, they make a ticket, and then the food comes back out to you. And I think in your vision of the restaurant future, the first step is pretty much completely gone. You're inputting your order as a customer, and then that automatically goes and is created as a kitchen ticket

with, like, less interference with no walking around, and then you have a waiter or a waiter to help with like special dietary restrictions or things you really need to talk to a human about. Yeah, I mean, I think there's different categories. Fine dining, I think for a very long time or maybe forever, will continue to be human first because you're paying for that experience of having the waiter or the waitress explain to you what the specials are and what's good and what's not good, et cetera. But that middle casual tier, I would posit, is ripe for disruption by this type of automation. You don't go to Ruby Tuesdays or, you know, I don't want to pick on anyone, but you don't go for the experience with their waiter-waitress. You go because they've got great food,

they've got great stuff on TV, it's comfortable seating, whatever. And you're often frustrated by how long it takes the waiter, the waitress, to get to your table, to order another drink, to get you the check, blah, blah, blah. That can all be fixed by voice technology. And you can still have a pleasant experience when the waiter, waitress brings you the food, you could have a chit-chat, sure, and you can keep it personal. But the companies that get this balance right are the ones that I think are going to succeed. Back to what you sat on Alexa, I want to make sure we spend some time there.

One of the interesting things about the voice technology hype and adoption cycle is that it seems like we're fairly far along on the adoption cycle. I'm comfortable using Siri. I'm afraid to say her name, but Alexa for the weather on a daily basis, but there isn't a ton of hype around what's going on with voice technology. And what's going on with that disconnect, do you think? Yeah. So voice like a lot of technology, and mobile was the same way, right?

Mobile technology started coming out in the late 90s with WAP and a bunch of other services. And everyone wanted this stuff on their phone. And it just, it was the same thing. You kind of checked your news, maybe, checked a couple emails, but that was annoying because he couldn't really respond. And it just, the interface wasn't right, but everyone knew that eventually we would have the internet or all of interactivity on our, on our devices. And what broke that open was, was when Steve Jobs ultimately launched the iPhone with a

whole new interface touch, et cetera. The same thing's going on with voice. We've been messing around with it for 10 years. We all want to use it because it's so fast to speak. We can speak three times faster that we can type, so it's just so efficient, but we don't want to listen. And the core mistake that all these systems have made, starting with Alexa and Siri, is that they try to create assistant human-to-human voice environments. They try to replicate those. But that's how what we want.

We just want to say something and have the machine do something or give us the information. And that's when voice will have the real interface breakthrough that it hasn't quite had yet. We all have it, but we don't use it for very much. Yet. And so that is the breakthrough. And by the way, Gen AI is unleashing all this because with Gen AI, voice gets so much better because the responses get better, but also the interpretation gets better. So I won't hold you to the prediction. But what do you think is that breakthrough moment then? As a consumer, all know that the breakthrough in voice technology has happened when? When basically every app you use has a giant mic button and you're primarily communicating with

your apps via voice first. and you're asking your bank account for new checks or to transfer money, and you're asking, you're doing all your orders, you're ordering your movie tickets, and kids are already doing this, right? In my house, nothing gets watched on the television that a kid hasn't yelled at the system to get them, right? They're not using fingers. They're using their voice. And I think that's one of the interesting conversations that you're having as a parent to talk about in the book is the way that essentially the pull, in the ways that we need to speak to the voice assistance.

And if it's even important to say, like, please and thank you. Yeah, I mean, one of the interesting things is the first generation of voice assistance were all female, right? And the reason was because they tested better. But the reason they tested better was because of human bias, right? Both men, and it turns out many women prefer to speak to a, especially to issue commands to a female-sounding voice bot, which is obviously massively problematic, and a lot of that is getting solved over time. But in our household, right, my 11-year-old son spends a lot of time

telling Google that Google is stupid. And it's, you know, it's something we're really trying to wrestle with. My wife is constantly correct. And the answer is, Mom, it's just a bot. It's a bot, but it doesn't feel like one, right? And you never know who's listening in. Maybe there's a person on the other end of it. We like thinking about top dogs, is investors. Who's winning? Who's really getting a lead in something? In Voice Tech, it seems like there's a couple, maybe at least on the surface obvious ones, even if you may disagree with it. It seems like Amazon might have won the smart speaker battle with Echo and Alexa. It seems like Apple might be in a heavy competition with Google for the voice assistant. That's more of maybe a

phone battle, but also with, I would say, microphones with AirPods. And I don't know, who are the top dogs maybe that you're watching or maybe who are some ones that we're not looking at? Yeah, so if I could break it down real quick, amongst the top players. So Amazon gets why voice is important. They want to control the home. That's why they've invested so much. I think they have a huge disadvantage because they don't have a screen that you have always on, always with you, which is your smartphone. And so they've invested a lot. I don't know that they're going to end up being one of the winners

because of the complexity around their, most of their experience are voice to voice, et cetera. Apple with Siri, that's probably my favorite chapter in the book is talking about the founding of Cherry, Siri, the interview with Adam Shire, his whole story of how they were acquired, and that Steve Jobs' vision for Siri has never really been realized, but I think it's about to be

because I think finally Apple is going to start, up their apps to developers so that you can start putting a voice layer very easily on every app. When that happens, this multimodal system will start to get unleashed. And I think Apple, because of that as an ecosystem, will be a big winner. Similarly with Alphabet, right, through everything they're doing Android and they have the in-home devices. The secret leader in all this, though, is Microsoft. And partially because Gen.A.I. is so important to it. And everything going on with OpenAI, but they also bought Nuance two, three years ago. And Nuance is the most developed voice transcription and recognition system in the world.

And Microsoft is going to combine those two. Now, they need to use, they're going to be able to license that to app developers, et cetera, so they will be a major, major player, but more in the background. And then meta, you know, one of the interesting things about, if meta is right about the metaverse, the metaverse goes through voice. You cannot interact with a metaverse with a keyboard. So meta is going to have to make massive, massive investments themselves. Now, they might partner with Microsoft or someone else, but they can't get to the metaverse without solving their voice issues. Those are the big ones. I think there's second order. I mean, as always,

Nvidia and other of the big chip players are in the background because one of the breakthroughs right now is that so much of the logic sits on your device versus in the cloud. and that makes it so much faster. So a lot of the chip players, and they're each going to have different chips that are coming out that address this market. And then there's some small, you know, standalone players like Soundhound, et cetera,

that are small public companies that are, as always high risk, but are all trying to solve a niche or get acquired that are also players in the space. The other thing I would look for is, as an investor, which companies are adopting voice and are they doing it well, right? And what I go back to is if you take the ordering of a movie ticket, right,

that takes two to three minutes right now in an app, in a voice experience where you ask your app, hey, Regal Cinemas, what's playing tonight? And it shows you on a screen, and then you say, get me two tickets to Star Wars at 8 p.m. You've not taken a two to three minute process into less than 20 seconds. And whatever we've learned on the Internet is that if you can take something from two minutes, 20 seconds,

you're going to win. And so I would be paying a lot of attention as to which companies are integrating voice experiences into their consumer experiences, into their customer support experiences, et cetera, because it will be a competitive advantage. So a lot to unpack there, and I'm not going to be able to get to all of it, I think. But I want to start then with Apple, because you tease the story a little bit. What was the vision for Steve Jobs with Cerell? and how is that being realized now? So initially, the idea behind Siri was not that it would be a voice assistant that only

does about five or ten things like it does today, but that it would be a layer that would be a communications layer that sits on top of the entire iPhone and on top of all of the apps. So that basically you can interact with any app using your voice. And Steve Jobs' big insight was, look, if it's on the screen already, you're just going to use your fingers to get there. But take a banking app. The average banking app in the U.S. has over 300 pieces of functionality. You can't organize that efficiently on screens that you slip, slide through, et cetera. The best way to get to something that you don't do every day, like reorder checks, like make a transfer, whatever it is, is via voice.

And that was the original idea. To date, even 10, 12 years after the launch of Siri, most of the application of Siri is still within the Apple app ecosystem. But I think that's really what's going to break open here soon. So, yeah, Apple's been pretty protective about licensing their technologies out. So Apple, I guess, is allowing Siri to be used with other apps. That's what you're saying? Well, I think it's going to be like their iOS development SDK.

So as developers, we get access to the iOS SDK. that allows us to build apps, they're going to now, I believe, start integrating voice so that just as easily as we have touch functions, et cetera, we're going to have a voice layer. All right. And then you also mentioned Microsoft. You said they're doing work in the background.

What is the work that Microsoft is doing with nuance in the background for voice tech? So some of this is not public, but if I were Microsoft, and I know that they're super focused on this and super smart about this, I would really be combining that the voice capabilities, which in the business is called conversational AI that comes out of nuance with the

generative AI that they have in the Open AI platform to make the voice experience is incredible. So when we speak to a, you know, to a Siri or to an Alexa, you know, according to Google, they get it right 98% of the time. I think most of us would say it's maybe 90% of the time. But what Gen.I is so good at is predicting the next word that you will say. I mean, that's ultimately the entire logic or the system. And so using Gen A.I., Microsoft can get their system so much better than others who are less adapted Gen AI in terms of interpreting what we're saying and getting the transcription right, which is the first thing you have to do in voice.

The second thing you have to do is then figure out what did the person mean. Again, Gen A.I is very good at that. And so given what they're doing with Open AI and there's other, you know, the Gen AIS, I think, now more open than we thought maybe it was six or 12 months ago. But Microsoft is really well positioned to then have a system that they licensed to app developers, web developers, et cetera, that said you can just add our voice layer that's powered by Gen A.I. to your experience, and that's the most efficient way to go to market. So rather than solving the problems individually of, I guess some of the programming challenges, which you've talked about,

I guess I'm curious which the largest one is. Microsoft has it. So figuring out the conversational parts where there's like 500 ways to order a cheeseburger, just use the open AI technology. We can pretty much cover you there versus you trying to go through and develop this technology on your own. Exactly right. I mean, that's what Microsoft is so good at, basically giving you an operating system now for voice. So what is the biggest challenge then? So voice tech has a lot of them, Accurate transcription, which I've seen get a lot better on one. We use the, it's called the Quarter app, and they have transcripts of earnings calls pretty immediately. And it's pretty darn close, quick to when the earnings call comes out. So accurate transcription, you also have

people talking with, I would say, bad microphones. So you might have garbled speech going into the system. You have different accents, and you also have the challenge of what someone is trying to say versus the parsing out the meaning of what someone is saying in these systems. Right now in voice technology, what do you think is the biggest challenge that developers are facing is they try to create more voice tech systems for their own apps? Yeah, I think it's translating the intent of what someone means into the action you want the system to do. And so, you know, again, if I'm ordering, if I'm ordering, if I'm ordering ordering anything and you mentioned cheeseburgers, there's, you know, Domino's has said,

there's over a billion combinations of pizza that you could theoretically put together. And these are very, in the grand scheme of life, very simple things, right, ordering a pizza. And so we still are struggling to get those 100% right. This is, again, why I believe that this concept of an overarching voice assistant is decades away because life is so complex. We got to get the pizza ordering right and the bank ordering right, et cetera. But these are solvable problems because they're bounded problems. The hard problems are the unbounded ones.

And so I think using generative AI combined with voice, we're going to make enormous progress in those areas over the next two to three years. And you're going to start seeing these voice experiences everywhere. I want to talk about voice and cybersecurity. I think there's a tie here, though. So one of the things that I think is probably the most entertaining out of voice tech and AI right now is like these music remixes coming out. And I don't know if you've heard this. Rick Engdahl, do you have the one with, it's Johnny Cash doing a cover of Taylor Swift.

Hello, I'm not Johnny Cash. So it's going to be forever or it's going to go down in flames. You can tell me when it's over. if the high was worth the pain. Got a long list of ex-lovers. They'll tell you I'm insane. Because you know I love the players and you love the game.

Tobias, you spend a lot of time writing about and thinking about voice technology. When you hear these clones, when you hear these like AI remixes, what do you think? A, it's super exciting that the technology has gotten this far and really quite quickly over the last 12 or 18 months. It is frightening, though, on the surface especially in terms of the security and the fraud that can be unleashed

there. I will say one thing. It sounds to us exactly like Johnny Cash. If you were to try crack into a voice-based authentication system, it's always an arms race because there are still fingerprints on the, you know, in the system that will tell some, not a human being, but a system that is designed to figure out, is this real or not? Today, they can figure out it's a never-ending cat and mouse game on the, on the security side. But what is most dangerous is not using this to crack into your voice password. What's most dangerous is how this can be used fraudulently to manipulate, especially the elderly that, hey, this is a call from your, I mean, you're in my voice today can be cloned.

And so based on this podcast, and so someone using this voice to call someone in our lives is a real danger. That's where the danger is going to happen is the human manipulation, just like the email scams that continue. But 10, 15 years ago, when it started, it was particularly pernicious and difficult to stop because no one had any idea this was coming. Yeah. So, I mean, with the cat and mouse game, though, I do think the voice cloning is in some

ways winning. So, Walsh's Journal tech writer Joanna Stern has a great video on YouTube where it's like, basically, can AI replace me for a day? And one of the things she's able to do is use a clone of her voice to get into her Chase Bank account. So there are financial, like, there's financial institutions where this is working or where one could hack into it this way, I think. And the worry that I would have is, I guess, why not just stick with two factors? authentication that like why do you need the voice technology in order to log in when two-factor authentication would clearly be safer in my opinion for a lot of these like financial bank accounts that financial accounts bank accounts what have you yeah i mean i think they're using two-factor in the

background because they're making sure that you're calling from a known number and they're combining that with voice but really we're probably going to three-factor where then you have to type in a four-digit pen Yeah. But I think we have to go to two to three factor authentication for sure. Single voice only from any from a random number. And I'd be interested on that chase thing. I suspect she's calling from her own mobile phone, right? I believe she did.

Yeah. So that might help with it. Yeah. It would not have worked if she were calling from another phone. But still, it obviously worked. As always, people on the program may have interests in the stock. they talk about. And The Motley Fool may have formal recommendations for or against, so don't buy ourselves stocks based solely on what you hear. I'm Mary Long. Thanks for listening. We'll see you

tomorrow.