Top news stories for Episode 12 (September 21, 2017):
1) A Vision Of The Future: Amazon is developing Alexa-powered glasses, possibly ready for market by the end of 2017.
2) Voicebot.ai Story Of The Week: VoiceLabs data indicates voice app "retention" has doubled in last nine months.
3) Amazon's Fire HD 10 is an "Echo Show in tablet form." Are tablets meant to be voice-first? Will this be a winning hardware play?
4) Google Home "Mini" to be announced in Google event on October 4 and will compete head-to-head against Amazon's Echo Dot.
5) Roku is working on a 'smart speaker' of its own. Does it have any hope of competing for significant market share at this point?
6) NPR interview, probably correctly, calls creating Alexa skills "working for Amazon for free."
This Week In Voice available via:
The Medium post referenced by Emily M. Bender in this episode, titled Google Home vs. Alexa: Two Simple User Experience Design Gestures That Delighted A Female User, is here.
Panel for Episode 12 (September 21, 2017):
Emily M. Bender is Professor of Linguistics at the University of Washington (her course on ethics in natural language processing is here), as well as Faculty Director of the Professional MS Program in Computational Linguistics (CLMS).
Named one of Business Insider's Top 75 Designers in Technology, Karen Kaushansky is an experienced product designer and experience designer specializing in voice user interface design.
Jess Thornhill is product manager for Life Bot, part of Opearlo, with years of experience working with emerging technologies including voice, AR, and VR.
Bradley Metrock: [00:00:10] Hi, and welcome back to This Week In Voice for Thursday, September 21, 2017.
Bradley Metrock: [00:00:18] This Week In Voice is brought to you by VoiceXP, blazing the trail in voice technology. VoiceXP is taking the lead in developing Alexa skills for some of the best brands in the world. With VoiceXP, all you have to do is say it to revolutionize your marketing strategy. And if you have not gone to www.VoiceXP.com, stop the podcast just for a minute...go there...check it out. You'll be glad that you did.
Bradley Metrock: [00:00:49] I'm Bradley Metrock - I'm CEO of a company called Score Publishing, based here in Nashville, Tennessee. We are very pleased to have a phenomenal panel today.
Bradley Metrock: [00:00:59] Our first guest is Professor Emily Bender - Emily, say hello!
Emily M. Bender: [00:01:03] Hello. I'm happy to be here.
Bradley Metrock: [00:01:05] Thank you for joining us. Emily Bender is a professor of linguistics at the University of Washington, along with faculty director of the professional M.S. program in computational linguistics. Emily, share with us just what is the study of linguistics?
Emily M. Bender: [00:01:22] So the study of linguistics is sometimes described as the science of human language. We're interested in how languages work, how they change over time, how they're similar to each other and different from each other, how people process language when they're talking, when they're understanding, how children learn their first language. how older people learn second languages...and Computational Linguistics has to do with how we can get computers to naturally deal with human language.
Bradley Metrock: [00:01:48] That's really cool. Thank you very, very much for joining us today.
Emily M. Bender: [00:01:51] Thank you.
Bradley Metrock: [00:01:52] Our next guest is Jess Thornhill. Jess, say hello!
Jess Thornhill: [00:01:55] Hi, thanks for having me.
Bradley Metrock: [00:01:56] Thank you for joining us once again. Jess, when we spoke last time, you had just switched on to Life Bot - I think Life Bot had just gone live. Share with us a little bit what you're working on right now, and a little bit about Life Bot.
Jess Thornhill: [00:02:08] Yes. Life Bot is a voice app, so it's available on Amazon Alexa and now recently, very soon hopefully, on Google Home. And the idea is that we want to create a voice app that will be the only app you ever need. And also help them do simple tasks and use Alexa and a voice assistant in the best way possible. So we're starting with productivity to help you do things like setting reminders and to do less...but the idea is that we want to become a platform where you can do almost anything.
Bradley Metrock: [00:02:40] Thank you very much for joining us, Jess. We appreciate you sharing your time and insight with us today.
Bradley Metrock: [00:02:45] And our third guest is Karen Kaushansky. Karen, say hello!
Karen Kaushansky: [00:02:48] Hi from Switzerland!
Bradley Metrock: [00:02:50] Karen...first of all, you can't say that and we can't explore that. So, why are you in Switzerland?
Karen Kaushansky: [00:02:56] Just moved out here with my family for a big adventure. But I am continuing to work in the field that I've been working in, which is working with emerging technologies: pushing the boundaries on things like voice, and conversational interactions, and self-driving cars.
Bradley Metrock: [00:03:14] Very cool. I did not know until doing research for this episode that you had been named one of Business Insider's Top 75 Designers in Technology. Congratulations on that.
Karen Kaushansky: [00:03:24] Thanks. That was a few years ago - I feel like they need to put out an updated list and let's see, you know, if I can kind of keep on it. But that was a few years ago already.
Bradley Metrock: [00:03:35] Yeah. Still very cool. Thank all three of you for joining us today. Thrilled to have you!
Bradley Metrock: [00:03:40] With that, let's get into the news. Story number one is a very interesting one, and what has become normal for this show...it involves Amazon. Amazon is developing Alexa-powered glasses, and if that weren't enough of a story, there's more! They're going to be ready by the end of the year. This is really going to shape a lot of how #VoiceFirst technology evolves moving forward. And Karen, I will start with you on this story: what do you think about Amazon trying again with glasses - you know Google Glass, they (Google) tried, didn't work. What do you think about Amazon trying again, and what did you take away from the story in general?
Karen Kaushansky: [00:04:22] So when I first read the headline, I got very excited. I'm currently working with a client working in mixed reality, and have been recognizing the power of voice in mixed reality...in, you know, virtual reality as you're augmenting the world around you and what you might want to do with voice. But then I dug into it a little more, and found Alexa - Amazon is creating this. They won't have a camera, supposedly. No camera and no screen. Basically, I think they're just trying to introduce Alexa in a way that you can take her with you, on the go, because they're not built directly into the smartphone, which we all know. But what I have to ask is what does Alexa know, or give me on the go, that other assistants don't? And is it worth wearing glasses for? I use Alexa to play music. I love my Alexa. I reorder toilet paper. But I don't know that I need to wear something extra in order to do that on the streets. I got excited, but then I was a little disappointed, because I'm not sure what the real use case is if it's not kind of pushing boundaries in other areas.
Emily M. Bender: [00:05:32] So I think I had the same question, which is it's interesting to think about where it might go. But if the point is all audio, then why glasses? I already wear glasses with corrective lenses - I don't want to switch them out for something that's audio. I think Amazon might do better to do something that's...I guess they want access to the the skull behind the ear, to be able to have it be really quiet? So you could do some sort of cool ear decoration or something, and leave the glasses out of it.
Jess Thornhill: [00:05:59] I think potentially, if Amazon has some kind of a longer strategy behind this, it could be quite a clever move. I think, obviously, this is probably their attempt to compete with Google and Apple in the fact that they don't have hardware - as in mobile hardware, that's on the go - so obviously you can access Google Assistant and Siri on your phones. Amazon has been behind the others, because you can't do that right now. So I guess this is their...and obviously the Amazon phones didn't do that well. But with something like this, it lets you access Alexa on the go. I agree that only being able to have audio doesn't really make sense for glasses, but it could just be a way of Amazon getting out a glasses product first...which, obviously, they seem to care about being first in the market with the Echo devices. Maybe this is just their way of getting something out there early, with the potential of having future versions that will allow for things like cameras and augmented reality, because there are rumors that Apple are working on AR glasses, and obviously you've got Google Glass. Maybe this is just their first iteration of it, and we'll see something more exciting in the future.
Karen Kaushansky: [00:07:13] Bradley, just thinking about a conversation you guys had a few weeks ago with the Microsoft Cortana and Amazon deal: the other thing that came to mind is are we going to now see people walking down the street with Amazon Glasses saying "Hey Cortana"? Right? All of a sudden, the lines are blurring, a lot. I just don't know what the glasses will do for the whole experience.
Bradley Metrock: [00:07:38] Well it's funny you say that, because I just got back from the Intelligent Assistants Conference out in San Francisco. Great conference, great organizers. Opus Research really did a nice job. Met a lot of folks out in the Bay Area. But I will say this: at the table that we were set up at in the vendors area, up walks some guy - I'm not going to say who it was. And explaining to him, you know, we do a bunch of shows on VoiceFirst.FM, including This Week In Voice, you know we're going to be recording that later this week...a lot of the stuff that is going on with the Alexa is moving at a fast pace. And he stopped me, and said "Yeah, Google's really doing a lot of great things with that, aren't they?" I say that only to say...not to call anybody out, but just everything's moving at such a fast pace, that this guy was a professional in this industry, and even he is confused between Alexa and Google Home. And you can just imagine the layperson is really starting to get buried under all of these product developments.
Bradley Metrock: [00:08:39] And to all of y'alls point, I think it's incumbent on Amazon to educate the market on how they would supposed to be using these glasses. But, speaking for strictly myself, if they do come out with one that is more than audio, that has like an Echo Show sort of screen built in in some sort of augmented reality type of way, man I am on board for that!
Karen Kaushansky: [00:09:02] That's why I got excited. I think that there is room for things like that, and maybe that will come. But clearly not...maybe not in the first rev.
Bradley Metrock: [00:09:13] So we will move on to story number two, which is our second VoiceBot.AI "Story of the Week." This week, it is the story that is based on some data that VoiceLabs has sourced indicating voice app retention has doubled in the last nine months, and Emily, with this I want to start with you. What did you take away from the story? And do you think...sort of the overall theme, I think, of the story, is that usage patterns of smart speakers and voice technology are shifting and evolving. Where do you think we're headed with all of this?
Emily M. Bender: [00:09:47] So clearly we're at the beginning of the curve here. I mean doubling is going from 3 percent to 6 percent - there's a lot of room for that to go higher. But I think the main obstacle for many of these apps is ease of use. And that ease of use, in particular, has to do with can the conversation be natural? And can the conversation continue for more than one turn? Right now, we're seeing a lot of the benefits of voice-based interfaces, but oftentimes you sort of can do one thing at a time, and then you have to say and wake word again. Or there's no saving the state of the conversation...what linguists would call 'the common ground,' or very rudimentary common ground. And especially with all of these skills that are being contributed through that platform that Amazon's making available, I think we're going to see even as some groups figure out how to do multi-turn conversations, a majority are not going to, and so it's going to be confusing for users for a while. But is this something I can keep talking to, or is it really just one-off commands?
Bradley Metrock: [00:10:49] But you do see it as a positive, right? That this number has gone up?
Emily M. Bender: [00:10:52] Sure. I mean it seems like, to a certain extent, developers are looking at how to create something that's engaging enough - and useful enough - that people find it worthwhile to use.
Jess Thornhill: [00:11:01] Yeah, so for us as builders of voice apps, it's obviously great news to see that more people are keeping on and using them, and in the second week. I think it definitely shows two things: so one, as Emily said, it means that hopefully this is showing that developers are creating better conversations. They're creating skills that are easier to use - that people know how to use, and are more likely to use, longer term. But also, I guess, it does show the general trend that people are starting to get more comfortable with voice, and their understanding of this can be a part of their life and something that they use every day. But I would definitely say that the fact that it's doubled is obviously a good thing. There's definitely a long way to go. We did a survey, with some of our users a couple weeks ago and asked "how many third-party skills they used on a regular basis," and a lot of them still don't know the difference between native Alexa and the third-party skills. So I think there's a massive amount of education that needs to happen on what a skill is, and how to use it, and how it's actually even different to what Alexa can do natively.
Karen Kaushansky: [00:12:04] Six percent's still pretty low, and I think what we'll talk about this with another story on quality versus quantity of skills. A lot of it is about discoverability and education. I'm curious ... what I'd love to see is the types of voice apps that are being used to understand how are people really using this? I mean we see the main ones, right? We're using it for music. We're using it for reminders. But what are those third-party apps that are really sticking? That are really the use case for these speakers in the home? And I'm also curious because, just recently, Alexa just launched not too long ago in Europe and it is very new for Europe, and the U.K., and in Germany and in France, and so I'd be curious to see also the voice apps across cultures, across countries - what's making a difference for people.
Bradley Metrock: [00:13:03] Emily, you touched on something interesting: the concept that you noted of 'common ground' in a conversation, and the fact that you're having to use the wake word every time. One of the big things I do with the Echo is play SiriusXM, and every single time, I have to say "Alexa, play '90's on 9' on SiriusXM," and if I say any of that wrong, or if it doesn't understand it, it just misses the whole thing. It would be nice to say "Alexa, keep playing music," you know, or additional context of the conversation, so I find it interesting you noted that. And I completely agree.
Emily M. Bender: [00:13:39] I just want to give you another fun example: we were heading down - I live in Seattle - down to Tacoma for the state chess tournament. And I was interacting, in that case, with a Google Home while I was, I think, brushing my hair, trying to think about "OK, where is the closest grocery store to where I'm going to be, so I can get snacks for my kid?" And I could not get it to give me the information. I would say where's the closest grocery store to the location that I was going to be at - the Tacoma Convention Center - and it was basically picking places close to me, where I am now, cause I said 'closest.' And then if I said OK, what's the address of the Tacoma Convention Center...and then in order to get it some of the stores near there, I had to tell it back that whole address. But if it had common ground, then I could say "OK, closest grocery store to that," and it would understand what I mean. That's the future I would like to see coming about...but those are really hard problems.
Bradley Metrock: [00:14:27] Yeah they are. And, you know, the flip-side of it is that if Alexa just stayed on all the time, then you'd have people screaming about privacy and things like that. So, yeah, we got...these are interesting problems that I'm glad I don't have to solve. I'm glad Amazon will handle that for us.
Bradley Metrock: [00:14:45] Moving on to story number three...and this is an interesting one. Amazon has rolled out a new tablet called Fire HD 10 that is, according to this review that's linked in the show notes, "Echo Show in tablet form." I find it very interesting because, as I've noted several times on this show and a couple of others, I am in love with the Echo Show. It's a phenomenal device - it's almost a magical device - and it's tough to even describe for folks who have not experienced it yet, or haven't used it...so anyway, this got my attention. Jess, I want to start with you on this...are tablets meant to be voice-first? Is Amazon going to win with this, like they have with the Echo Show? What are your thoughts?
Jess Thornhill: [00:15:26] Yeah, so I found this really interesting as well. And of our users, a lot of them do actually already use Alexa on their Fire tablets, which we found really interesting because at first we just assumed that 100 percent of people will probably only access Alexa on an Echo device. So it's really interesting to see that tablets coming into the fore here. And if you can give someone almost the same capabilities as an Echo Show in a much cheaper, more convenient format, then I say go for it.
Jess Thornhill: [00:15:57] The one thing I think that was interesting on this, the way that this article was written, was that it kind of suggested that it was counterintuitive to have voice on a tablet, because it sort of suggests that why would you need to use voice when you can use your fingers, or your hands? And I think that's one thing about the discussion around voice I think needs to start changing - the idea that you only use voice when you can't access your phone or your tablet. I think it's not necessarily the case. If anything, voice should be considered as something that is complementary to, say if you're browsing on your tablet, but you're doing one thing but you want to also complete something else - another task at the same time. You should be able do you that with your voice, and I think using Alexa and tablets is the way to do that. I think it's important to start thinking about voice as not just a sort of secondary alternative to using a mobile device.
Bradley Metrock: [00:16:55] What I immediately think about with this is education. With other parts of Score Publishing, we've gotten involved in producing content in the educational realm. And Emily, I want to ask you...you teach linguistics. You teach a number of classes. Wouldn't it be cool to have a textbook - a digital textbook or other digital resources, that are part of your curriculum, that are on a tablet device that could be voice-first? I just think there's so much potential for that.
Emily M. Bender: [00:17:25] That's a really interesting question. I think that there's, as Jess was saying...it's worth thinking about how we use voice naturally, and how it could fit in. A lot of what we do with all kinds of devices is very much shaped by how the devices have trained us, so it's really hard to get people to do full natural language queries in internet search because we're so trained to the keyword thing. Even if we have something that's hard to express in keywords, we still try. And so figuring out where the natural uses for voice are, and where it isn't, is not convenient. Right? So one thing about your hands is that they're quiet. And so there's sometimes when you've got to be quiet, and so talking wouldn't be good. And then there's sometimes where talking is the more natural alternative. And so I think that it could be, for example, if I had a textbook on a voice-activated or voice-capable system, I might want to be able to search in it - maybe search for terms I don't really know how to spell. So that's one side of it - sort of finding the use cases where it makes a lot of sense.
Emily M. Bender: [00:18:22] And then the other side is an issue that I think applies across voice technology, which is being ready to deal with diverse speaking styles, diverse dialects, accents...so thinking about my own students, we have quite a few international students, and I want to make sure that if I integrate them into my curriculum, it is equally accessible to someone who's English doesn't sound like mine.
Karen Kaushansky: [00:18:42] I am curious to see what the performance is like. I think one of the things that Amazon broke out into the market was their hardware - their mic array, the ability to talk to Alexa from across the room. And this is different, of course, because you have a screen, and when you have a screen, you're kind of expecting that you're going to be closer. So maybe it will still pick up "Hey, Alexa" from a little ways away and it'll work.
Karen Kaushansky: [00:19:12] But the one thing that I wanted to bring up as well with voice on tablets, or you know, it's not just voice, or gesture, or touch...it's that we're really going to start seeing situations where you're going to want to use both. Which is, you know, the old MIT "put that there." And "I want that one." That's where, again, kind of more where the cameras kind of integrate into these devices - are going to become kind of required. But, I think using both voice and touch and gesture, all together, will end up making the experience much more natural. So it's not only common ground with voice - it's like common ground with context as well.
Emily M. Bender: [00:19:56] I just want to chime in and say absolutely, that the common ground is not just the linguistic history - what's been going on - but the environment that the interlocutors are in, and what they can see, and how can they can direct each other's attention to things, with things like points. I think that's spot on.
Bradley Metrock: [00:20:09] Moving on to story number four: Google is rolling out a mini...a Google Home Mini that looks similar to the Amazon's Echo Dot. Apparently it's designed to function like Amazon's Echo Dot. They're announcing this thing on October 4th, according to the report that we have linked in the news stories of the week. And Karen, I'm gonna start with you on this. You know, we talked about Amazon - it's dominated this show, it's dominated other shows. Google, clearly with this, simply copying and pasting Amazon's approach. Is that going to be enough? Or is there something else that Google ought to be doing to compete in this space? What did you take away?
Karen Kaushansky: [00:20:52] Well I think there's still a lot of room in this space. I mean, it's not a surprise. They want more entry-level speaker in people's homes, so they want to compete. But I think they're also trying to get into as many homes as possible before a Facebook speaker comes or other, you know, the Apple speaker comes. So how do they get in more people's homes, get more data, get better over time? I mean this is what Google does really well, which is improve over time, right? It's all about the data, and making it better. So I think they're definitely not - and this isn't a surprise - and I think I think they're doing the right things.
Bradley Metrock: [00:21:39] Jess, with what Lifebot and Opearlo are doing, does this excite you? Google coming out with their own version of the Amazon Echo Dot? So they can get more and more penetration into the marketplace? Or do you think they ought to be doing something different?
Jess Thornhill: [00:21:53] I think they should be doing this, and to be honest, I'm quite surprised it took them so long to come out with this smaller version, because the Echo Dot's been around for quite a while now. So I think this is definitely something that needed to happen. I don't think it's particularly revolutionary. And I think it's just another step that they have to do, seeing as Amazon already released their own version.
Karen Kaushansky: [00:22:18] I think we're also going to probably see, you know, multi-room audio. Right? So now you can have your big one in one room, and you're going to get a mini for the other room. Amazon is doing it already. One thing that Google is doing, that Amazon isn't, is the speaker identification. So I mean Google is pushing on innovation in ways that Amazon isn't, and I think they'll continue to find ways like that to push forward.
Bradley Metrock: [00:22:47] Story number five: Roku is working on a smart speaker of its own. And this has been sort of a recurring theme on this show: hey, here's another company doing a smart speaker...oh, there's another one! And perhaps that's a cynical view, or maybe it's not. And Jess, I will start with you on this: what do you take away looking at this, and is this a positive? Or is this just sort of irrelevant?
Jess Thornhill: [00:23:13] So I'm not too familiar with Roku - I'm not sure how big they are in the UK right now. But, I do agree with you. It just seems like every week there's another brand or company announcing that they're probably going to come out with their own smart speaker.
Jess Thornhill: [00:23:26] And I think it's good in the sense that it just means that more and more people will be using voice as a means of interaction - in a new way of communicating with technology. But I think the real question is now how are we going to provide a consistent experience across all these different platforms? We've already got Google Home and Alexa, but how are apps going to really provide an experience that is consistent? And it's not just about having the same functionality on your Google action as your Alexa skill. But say, for example, we get the stage where people are using different voice platforms in different situations. Say you might have Alexa in your home, but Cortana in your car, or Roku on your TV. How do you start an experience in one, but continue it on another? And I think that's going to be the real question, because, as you said your story about the conference where someone still said that it was Google and seemed to be a bit confused about who the players are in this field, I think it's going to be really important that people are educated and they know how to use voice and how to use it across different platforms.
Karen Kaushansky: [00:24:40] But Jess, are you suggesting that it should be consistent, or just that you can kind of take your experience on one assistant to another assistant?
Jess Thornhill: [00:24:49] I think consistency in a sense that, say for example, you had in iOS app, but then you switch to Android, you want to access the same features on Android. But I think the most important thing is this being able to carry your experience across different ones, because I do think that we'll get to the stage where people have different assistants in different parts of their life.
Karen Kaushansky: [00:25:08] I know that I was pinged for that voice user interface job, so I know Roku has...they've been trying to hire for a long time. I was probably pinged 8 or 9 months ago, so they've had these open for a while.
Karen Kaushansky: [00:25:20] You know, maybe they're making a speaker. Definitely, they're behind in adding voice as a really natural way to interact - to look, you know, to search for your TV content for what you want to watch. But again, it goes back to what you do well? What are your core competencies? And why am I going to go to a Roku speaker? There has to be some reason, some benefit that it offers me, over the ones that I already have...because you're right, Jess is right, I already have two or three assistants in my life, so what is Roku going to bring to me that's different?
Bradley Metrock: [00:25:57] Yeah, it's incredible audacity, isn't it? It's incredible audacity, and I noted this - this is the way I think about, like, Facebook saying they're going to come out with a smart speaker. The incredible audacity of a Facebook, or a Roku, or anybody, to create a device that they just assume that someone's going to put in their house, and perhaps listen to every word that they ever say, for the rest of their life, without a compelling use-case scenario. You know, Amazon is out there marketing every day of the week on TV: "here's what you do with this thing." But most of these other ones aren't - it's just "hey, we'll throw this out there, and watch as you ring it up, and take it back home, and keep it in your house." I agree, it's definitely saturation.
Bradley Metrock: [00:26:46] So, what I thought - and this is again building off of your story from the person who had the Google Assistant and Alexa mixed up - my advice to Roku and to Facebook and whoever else is just jumping in, is to think carefully about how you name the assistant. I think that Google has done well by, when you talk to the thing, you say "Hey Google," so you know what you're interacting with. I think Roku should do the same, both for this practical sort of brand identification, but also the context that we've been talking about. So that, if there are differences, before the skills become more consistent across them, it's easier for users to keep track of what they can expect from one that they're talking to...but also for important social reasons. So this put me in mind of a Medium post that I read, actually as part of the seminar I taught on ethics and natural language processing. It's called "Google Home Versus Alexa: Two Simple User Experience Design Gestures That Delighted A Female User" by Johna Paulino, from January of this year. And this is all connected with: what does it mean to assign a gender to these assistants, that are not people? And what does it mean to even personify them to that extent? And so, I think that Google has done a much better thing where you are talking to Google - effectively, the company, or the software - where Amazon has created Alexa, given a feminine-sounding name, people refer to it as 'she' ... and that leads to sort of icky interactions, as Paulino explains in this Medium post.
Bradley Metrock: [00:28:11] I would love to see that Medium post, by the way, and we will link to that in the show notes. Would love to include that.
Bradley Metrock: [00:28:16] So let me follow up on that: so what is considered the best practice? Or the theoretical best practice, for an intelligent assistant, or a voice assistant? It should be genderless? Or male in some situations, female in others? Or what was the takeaway?
Emily M. Bender: [00:28:33] So my takeaway - and I'm not sure this is the consensus - is that my takeaway is that you're better off leaving it genderless, and having it be an entity, an interlocutor, a creation...but not something that pretends to be a person. And if we get to a point where there's lots and lots of these, and then you might be able to say "OK, we're going to have some of them be male, some of them be female, some of them be other, and see that they all behave the same way," that would be positive. But right now, the trend is they are all, if gendered, they are gendered as female, and that can be quite problematic.
Karen Kaushansky: [00:29:05] I know that Microsoft did a bunch of research before they really launched Cortana. And they tried to make a very neutral-kind-of-sounding assistant. And then what they found was the stronger the personality, with Cortana, it was polarizing for some people, but... and kind of turned some some people off...but in more cases, it created more of a connection to people for re-use. So, even though it wasn't neutral for everybody - so, again, it did create some backlash for some people - it did create deeper connections to others. So I think that there is a very...a long debate about this, in the industry, of having a persona. Because we have been in the industry for a long time. We say that even if you don't design a persona, there is still one there. So you are talking to Google, the company...which is like information, and just trying to get, you know, trying search-based kind of interactions. But there is a persona there.
Emily M. Bender: [00:30:12] Yeah, I think we can separate the question of personality from the question of gender. So there was a company called Ozlo building a personal assistant, which I think was text-based and not voice, which has been acquired by Facebook. So that might be part of where Facebook is going. I don't have inside information beyond that. But what I know is that that was not a person. So it had a personality - it was a furry creature...
Karen Kaushansky: [00:30:34] I didn't know they got bought by Facebook. I do know Ozlo. Yep, he was a cute little character.
Emily M. Bender: [00:30:41] Yeah. I think that's an interesting way to go: can still be a personality, without setting up a situation where you're constantly telling this female person in your household, ordering it around. Or a female person-like entity, in your household, ordering it around, in front of your kids.
Jess Thornhill: [00:30:55] That was a great article in The New York Times, quite a while ago, about people who have become almost emotionally attached to their Alexa. And they have sort of personified it. And I think it is really important that your voice assistant has a personality, but obviously, it's where to draw the line. And I think there's a lot of debate about what the best practices are for that. Something we're doing at the moment, which we're finding quite interesting, is we're building our skill for Google - so we're building a Google action - and the Google, you can actually change the voice that speaks. And there's two female options, and two male options. We're still discussing this, because we want our app to have its own personality, as a sort-of sub-assistant under Google, but we want to make sure that we get it right, and that the personality we portray matches the voice options that are available to us. So we'll see how that goes.
Bradley Metrock: [00:31:50] This is an interesting conversation. This is a fascinating conversation. And, from my standpoint, with Alexa, I interact with Alexa every day. I don't think about that it's a female. I don't know...I'm sure I'm probably in the minority. I'm sure that probably makes a big impact. But what I will say is when I was at the Intelligent Assistants Conference, just a few days ago, I had a chance to meet Brian Roemmele in person. He's been a frequent guest on our show - just a brilliant individual. And he does a lot of work creating his own voice-first prototypes, and doing independent research. And one thing I know that he's been working on is what it looks like to integrate Myers-Briggs into the development of an intelligent assistant, or a voice assistant. To me, I think that's absolutely paramount, because the gender can be one thing. You know, I'll talk all day to a genderless voice assistant, or a male, or female, but if that thing is chirping at me in a way that, like, it becomes grating, that's going to be a problem.
Bradley Metrock: [00:33:01] Moving on to story number six: there was a recent NPR interview - it's a brief one, it wasn't very long - and it is sort of retreads some old ground, in terms of Amazon and Alexa development. But one thing that really caught my attention, that I thought was pretty negative, is the assertion - that I think is correct - that creating an Alexa skill, for the most part, is doing work for Amazon for free.
Bradley Metrock: [00:33:26] So right now, just to recap: you develop an Alexa skill. Amazon's been great: they've developed the market, they've been out in front, marketing. They've really provided a lot of leadership in the marketplace.
Bradley Metrock: [00:33:39] But, what they haven't done is allow people to use the developer tools that they've provided to create voice applications that you can sell. Instead, they tell you, the developer, that you do well enough - and we're not even though tell you what that means - if you have a certain amount of engagement, we will send you a check. If you do not have a certain amount of engagement, we will not send you a check. And I think it's totally unsatisfactory. I've talked about this several times on the podcast before. I think it's one of the few negative things that Amazon has not addressed yet. And I just want to get all three of y'all, and Emily, I'm going to start with you on this. First of all, do you agree with me? Do you think this is a bad thing? Second of all, what other takeaways, if any, did you have from this piece?
Emily M. Bender: [00:34:29] Yes, I absolutely agree with you, and I'm very much worried about tech companies exploiting especially students, who are studying computer science and related fields, and saying "yeah, come to our hackathon," and basically getting them to work for them for free. And I think it's an important role that we play as faculty in educating students in the value that their skills hold, and to really say to them, you know, "think carefully about this. What are you getting for your effort? And is there another way for you to get that.
Emily M. Bender: [00:34:57] Similarly, I will sometimes be approached by employers who want to see if any of our students are available for unpaid internships. And my answer is always "well, you're welcome to advertise that, but since they tend to get paid internships, probably no one's going to take you up on it."
Bradley Metrock: [00:35:11] Yeah. That's a great example. There's so much discussion today about manipulation of people, and manipulation of labor, and stuff like that...it just makes it an even harder pill to swallow. So I think those examples were great, Emily. Jess and Karen, your thoughts on this? Either one of y'all...
Jess Thornhill: [00:35:30] Yeah, so I think we spoke about this the last time I was on this show, because they had just announced that they'd start rewarding more categories, apart from games, I think. And I don't think anything's changed since then. And, so now, they're rewarding people who build apps in productivity, and that includes Lifebot. But completely, as you say, we have no idea how they decide what makes this good, popular app - no idea how they decide how much you receive. And the money can change every month. And I think, especially as a startup, it is quite frustrating to have to sort of wait on Amazon to understand how we can start to really monetize. And there seems to be no real indication of what they're doing, and what their plans are on that front, and I think especially...I know a lot of young developers as well who are encouraged by the idea of potentially getting some money to just churn out skill after skill. And, in that sense, the article, even though it's very negative, it is right.
Jess Thornhill: [00:36:34] And a lot of the skills that are now in the store are of questionable quality, in terms of whether it's something that people actually want to use, more than once, and as more than just a gimmick. And I think as soon as there's a proper way to monetize, then it will encourage people to also develop skills that have more meaning, that have a better experience for users, and that's definitely something we're trying to, aside from the monetization issue, it's definitely something we're trying to really work on is to actually build a skill that people value, and that people want to use. So we're talking to our users every day. We're really trying to understand what makes a quality voice app. And I think that is the other side of what this article is saying: that the unfortunate side of this is that it looks like quantity is being favored by Amazon over quality right now.
Karen Kaushansky: [00:37:28] I love that idea, Jess. I hadn't really thought about it that way. If we had a good way to monetize, then we would be building better apps. So maybe it's a bit of a chicken and egg thing. But clearly now, I mean, there's just too many in there. They're pretty bad, or they don't work. I mean, I thought I found...I had a use case, and I really, we really tried to use it. It was a translator. It was a French translator - so give it an English word, and it would tell us what the French word was. And three - probably one out of every four times we use it - it couldn't...I would open, can't remember what it was called, the French Translator...it was down, like, three...one out of every four times we used it. So it was like "forget it, I'm not gonna, you know, to be disappointed twenty-five percent of the time, cause I can't access it." It's not worth it. If you don't care about you're trying to provide me a skill, you're trying to provide an experience, but you don't care about keeping it alive, then it's not worth it. So I love that idea. Let's make sure that they monetize properly, and we'll get better quality out of it.
Jess Thornhill: [00:38:43] And I think adding onto that...Amazon also have a responsibility to educate developers on what to do after they've published their app. I think they've got this massive push right now to just get people to get skills out there, no matter whether they even work properly, as you said. We've encountered so many skills that you can't even get onto, that don't work as they say they would. And I think one thing that many skills aren't doing is actually upkeeping their app, or updating it or fixing bugs that people are noticing. So I think it's one thing to help. Obviously, Amazon wanted a store - a skill store with a lot of skills that looks attractive for people to use. But I think they really need to start also providing more education for developers on what to do after it's been published.
Bradley Metrock: [00:39:30] Completely agree, across the board, and I think...you know, to the extent that the Apple App Store is an appropriate comparison for this - which I think it's a pretty good comparison for this, at least at first, for what's going on and what Amazon ought to be trying to do - when a app developer spends a lot of time...I know this, my brother is an app developer; he's based in Atlanta. When you spend a lot of time developing an application, you pour your heart into it, you pour your soul into it, and the intention is that you're going to sell that...you have the ability to sell that, you're going to sell that, and you're going to put food on your table with it. What happens?
Bradley Metrock: [00:40:13] Well, I can tell you what happens. When people respond to you positively or negatively, you are extraordinarily responsive. When somebody presents you with a problem, you fix it, immediately. When there's any sort of...when an opportunity for customer service arises, you provide it. It lifts the whole ecosystem up, just with the sheer fact that you could sell your product. And in a situation where you can't...this weird thing that Amazon's got going which only has one other comparable, and it's Amazon's other ecosystem which is in the publishing realm, and what they do with Kindle...this is an exact mirror of what they do over there. It just muddies the water - it leads to situations like Voicebot.ai reported on last week, which is that 62 percent of Alexa skills today do not have any user reviews. None. So I think an appropriate takeaway from that is, you know, people are just downloading stuff as a gimmick or not at all. And then, if it works...chances are, it doesn't work, or there's always going to be feedback to provide. But there's no impetus on the user to provide that - who cares? It's just all throwaway stuff. Hopefully Amazon sees the light on this, and allows people to monetize. And I'm hearing consensus from the panel on this. Any closing thoughts on this news story?
Emily M. Bender: [00:41:46] I absolutely agree. I think it ties back to that growth from 3 percent to 6 percent of people using it in the second week. I think that if we did have this scenario that Jess and Karen are talking about, where the app developers were more directly connected to their customers, and actually seeing gains for doing better apps, those numbers would go way up as well.
Bradley Metrock: [00:42:05] Sure. And we didn't even discuss...you know, all this story is about, or all we've talked about, is the ability to monetize the initial sale of an app. We aren't even touching on in-app purchasing or other monetization schemes that can build off of that. So anyway, I think the point is made. Hopefully Amazon is listening, if not to this podcast, then to their other customers, who I'm sure will be providing the same feedback, and maybe we get this at some point.
Bradley Metrock: [00:42:35] This was fantastic. Emily, Karen, Jess...thank you very much for joining us today.
Emily M. Bender: [00:42:41] Thank you for having us on. This was a lot of fun.
Bradley Metrock: [00:42:43] Thank all three of you for sharing your time, sharing your experience, your expertise, and your insight, with not just me, but the audience. It is greatly appreciated.
Bradley Metrock: [00:42:54] For This Week In Voice, Thursday, September 21, 2017...thank you for listening, and until next time.