WEBVTT - AI Assistants and You 0:00:04.240 --> 0:00:07.560 Get in text of technology with text stuff from hastaff 0:00:07.600 --> 0:00:13.520 what dot com. Hey there, and welcome to tech Stuff. 0:00:13.560 --> 0:00:17.520 I am your host, Jonathan Strickland, and today I want 0:00:17.520 --> 0:00:22.439 to talk about war. War never changes. But I'm not 0:00:22.440 --> 0:00:25.319 talking about Fallout, even though I just recently started to 0:00:25.360 --> 0:00:28.400 play Fallout for again because I never finished it. No, 0:00:28.480 --> 0:00:32.720 I'm gonna talk about a war going on in artificial intelligence. 0:00:33.240 --> 0:00:35.800 And it's not sky Net, it's not the Terminator, It's 0:00:35.880 --> 0:00:39.440 nothing like that. No, it's really about who gets to 0:00:39.560 --> 0:00:42.680 be your personal assistant. That's the best kind of war 0:00:42.800 --> 0:00:45.680 because we win no matter who wins, right, we get 0:00:45.720 --> 0:00:50.000 the best out of whichever combatants into the AI personal 0:00:50.000 --> 0:00:53.360 assistant thunderdome. I'm throwing out a lot of references to 0:00:53.400 --> 0:00:56.840 sci fi here. I'm gonna cut that out. So we 0:00:56.880 --> 0:01:01.000 want to talk today. We being me and you you 0:01:01.040 --> 0:01:02.760 can talk back. I just won't be able to hear 0:01:02.800 --> 0:01:07.480 you about these personal digital assistants, but not the p 0:01:07.600 --> 0:01:09.480 d as of the past. We want to talk about 0:01:09.560 --> 0:01:12.640 the the series and the Cortanas and the Google Assistance 0:01:12.680 --> 0:01:16.160 and things of that nature. And I want to specifically 0:01:16.240 --> 0:01:19.119 look into how are these going to be incorporated into 0:01:19.160 --> 0:01:21.399 our lives in the future, and what are some of 0:01:21.400 --> 0:01:25.400 the concerns we have and what differentiates all these products 0:01:25.400 --> 0:01:28.479 that have been sort of coming into their own over 0:01:28.520 --> 0:01:31.959 the past few years. So to start with, you might say, well, 0:01:32.160 --> 0:01:37.680 you know which of these assistants came first, And arguably 0:01:37.680 --> 0:01:41.440 you could say Google actually beat everyone to the punch 0:01:41.520 --> 0:01:44.759 by a couple of months because on June four, two eleven, 0:01:45.040 --> 0:01:49.040 Google announced at an inside Google Search event that it 0:01:49.120 --> 0:01:53.400 was going to roll out voice search on Google dot Com. 0:01:53.440 --> 0:01:58.680 And the project name at Google was uh Majel or Magel, 0:01:59.040 --> 0:02:01.120 depending upon how you want to pronounce it, but Majel 0:02:01.120 --> 0:02:04.480 would be the way her name was actually pronounced. Named 0:02:04.480 --> 0:02:07.520 after Majel Barrett, who was the wife of Gene Roddenberry, 0:02:07.560 --> 0:02:10.840 the creator of Star Trek. Majel Barrett actually played the 0:02:10.919 --> 0:02:14.040 voice of the computer system, particularly on Star Trek the 0:02:14.040 --> 0:02:16.280 next generation. Whenever you heard the computer speak, that was 0:02:16.320 --> 0:02:21.080 Majel Barrett's voice. She also played uh Deanna Troy's mother, 0:02:21.200 --> 0:02:27.480 Luaxanna Troy. Anyway, they named it after her. Internally, uh 0:02:27.680 --> 0:02:31.079 it actually doesn't have a name name, which kind of 0:02:31.120 --> 0:02:33.919 sets it apart from some of the competitors. So the 0:02:33.960 --> 0:02:38.120 Voice Command project was a tool from Google Labs, so 0:02:38.160 --> 0:02:42.640 their research and development arm and on March fourteen, this 0:02:42.720 --> 0:02:47.520 particular feature was rolled into the Google Now product and 0:02:47.960 --> 0:02:50.560 uh it was part of the Android four point one 0:02:50.680 --> 0:02:54.560 release that was the jelly Bean release. Now, at that point, 0:02:54.639 --> 0:02:57.640 the speech recognition commands had evolved a little bit. It 0:02:57.680 --> 0:03:00.560 had gone beyond some of the initial old stuff where 0:03:00.560 --> 0:03:02.960 you could just ask Google to search something for you. 0:03:03.040 --> 0:03:06.560 This was also a feature that was worked into Google Glass, 0:03:07.000 --> 0:03:09.280 so if you had a pair of Google Glass, you 0:03:09.320 --> 0:03:12.520 know that the voice command would always start with the 0:03:12.560 --> 0:03:16.919 phrase okay, followed by Google I'm not gonna say it together, 0:03:17.400 --> 0:03:19.280 just in case some of you are listening to your 0:03:19.320 --> 0:03:22.080 devices or listening with a device nearby and it's on 0:03:22.160 --> 0:03:24.320 its home screen. I don't want to activate it for 0:03:24.400 --> 0:03:28.040 whatever reason, but you can use that phrase that would 0:03:28.240 --> 0:03:31.800 end up alerting the virtual assistant that you wanted something, 0:03:31.840 --> 0:03:34.200 and then you would speak whatever it was you wanted. 0:03:34.240 --> 0:03:37.720 And over time, functionality increase, so it went beyond just 0:03:37.800 --> 0:03:41.280 searches and into more interactive features like with an Android 0:03:41.280 --> 0:03:44.560 phone you could set an alarm, or you could set 0:03:44.560 --> 0:03:48.600 a reminder or review your calendar and more. As time 0:03:48.640 --> 0:03:53.200 went on, at this point it is evolved into something 0:03:53.240 --> 0:03:55.840 a little bit more robust than that. Even you can 0:03:55.880 --> 0:03:58.360 start to interact with some third party stuff as well. 0:03:58.720 --> 0:04:02.560 And at Google Io two thousand and sixteen, it became 0:04:02.720 --> 0:04:08.240 part of Google Assistant. Now Google Assistant is really the 0:04:08.320 --> 0:04:13.240 intelligent personal assistant product from Google. The earlier versions you 0:04:13.240 --> 0:04:17.039 could think of a sort of a rudimentary form or 0:04:17.120 --> 0:04:20.200 or perhaps a prototype or maybe just like these are 0:04:20.240 --> 0:04:24.320 features that would eventually be rolled all into one finished product, 0:04:24.640 --> 0:04:28.160 being Google Assistant. So by that argument, if you say 0:04:28.200 --> 0:04:31.360 Google Assistant, you know, if you mark the Google I 0:04:32.279 --> 0:04:35.800 event as its premier, then it's not the oldest. But 0:04:36.080 --> 0:04:40.680 it dates back to June fourteen, two thousand eleven, when 0:04:40.680 --> 0:04:46.479 Google announced this initial search voice search ability. So that 0:04:46.600 --> 0:04:51.159 same year, in October, on October fourteenth, in fact, Apple 0:04:51.279 --> 0:04:55.560 introduced Sirie. And I'm sure you all know what Siri is, 0:04:55.600 --> 0:04:57.960 but just in case you don't, it's billed as an 0:04:57.960 --> 0:05:01.160 intelligent personal assistant and it was introduced as a feature 0:05:01.680 --> 0:05:03.920 with the iPhone for S and it's been part of 0:05:03.960 --> 0:05:08.119 the iPhone iOS ecosystem ever since. And it uses speech 0:05:08.120 --> 0:05:11.120 recognition to interpret user requests and responds with what is 0:05:11.160 --> 0:05:16.839 hopefully inappropriate action. According to series creators, Apple actually scaled 0:05:16.960 --> 0:05:19.480 back what Siri was supposed to be able to do. 0:05:20.200 --> 0:05:22.919 They said that they had arranged for Siri to work 0:05:23.120 --> 0:05:27.920 with about forty to forty five different apps that Apple had, 0:05:28.400 --> 0:05:32.520 and then the company scaled that back significantly, so the 0:05:32.720 --> 0:05:37.120 serie creators essentially sold the product to Apple. Then they 0:05:37.120 --> 0:05:41.000 went on to create a different intelligent assistant called VIV 0:05:41.400 --> 0:05:45.360 v I V and VIV is currently unaffiliated with any 0:05:45.400 --> 0:05:47.760 other big names, but it has received funding from some 0:05:47.920 --> 0:05:50.839 very wealthy folks in the text sphere, like Mark Zuckerberg, 0:05:50.880 --> 0:05:55.440 for example. And VIV is what the creators of vivs 0:05:55.480 --> 0:05:57.560 say that it's what Siri was supposed to be from 0:05:57.600 --> 0:05:59.960 the get go, and and essentially they're saying that Siri, 0:06:00.040 --> 0:06:03.000 you was kind of hampered, hamstrung, if you will, by 0:06:03.080 --> 0:06:06.120 Apple um And we'll get into more about why that 0:06:06.200 --> 0:06:10.240 may be in a little bit. So Sirie actually came 0:06:10.320 --> 0:06:14.760 second after Google had announced their voice search, keeping in 0:06:14.800 --> 0:06:17.480 mind that Siri was a different presentation, So you could 0:06:17.560 --> 0:06:20.400 argue that Siri was really more of the first assistant, 0:06:21.360 --> 0:06:25.680 and that the Google approach eventually evolved into an assistant. 0:06:26.040 --> 0:06:31.440 But wasn't really at that same level back in Moving 0:06:31.480 --> 0:06:34.240 forward in spring two thousand and fourteen, that's when Microsoft 0:06:34.279 --> 0:06:37.880 got into the game by unveiling Cortana, which is their 0:06:37.880 --> 0:06:42.239 intelligent assistant for the Windows Phone platform. And in twenty fifteen, 0:06:42.279 --> 0:06:45.560 Microsoft included Cortana with Windows ten. So if you have 0:06:45.560 --> 0:06:48.640 a Windows ten machine, Cortana is part of that, and 0:06:48.760 --> 0:06:51.200 if you have a microphone you can actually give voice 0:06:51.200 --> 0:06:54.839 commands to Cortana. You can also interact via text um. 0:06:55.080 --> 0:06:58.240 Cortana is named after the AI and the Halo franchise 0:06:58.279 --> 0:07:01.000 and is voiced by the same act Risks who provided 0:07:01.000 --> 0:07:02.880 the voice of Cortana in the games. So you can 0:07:02.920 --> 0:07:05.400 ask fun things about Master Chief and she always has 0:07:05.400 --> 0:07:10.560 a interesting answer for those. All of these, by the way, 0:07:10.760 --> 0:07:13.640 tend to have some sort of fun element to them, 0:07:13.680 --> 0:07:18.640 where the developers clearly thought of ridiculous things you could 0:07:18.720 --> 0:07:23.920 ask the digital assistance and built in responses that were humorous. 0:07:24.080 --> 0:07:26.640 For example, the big one that everyone talked about with 0:07:26.640 --> 0:07:29.040 Siri was where can I hide a body? And Syrie 0:07:29.080 --> 0:07:33.560 would come back with nearby quarries and caves systems and 0:07:33.600 --> 0:07:37.480 things of that nature. Now, in November two thousand fourteen, 0:07:38.640 --> 0:07:42.840 we get our final big name in this battle, Amazon. 0:07:43.040 --> 0:07:45.800 That's when Amazon unveiled the Echo, which is that sort 0:07:45.840 --> 0:07:50.600 of standalone speaker system UH that has the Intelligent Assistant 0:07:50.640 --> 0:07:54.440 Alexa incorporated into it. And like the other ones I've 0:07:54.440 --> 0:07:57.080 mentioned so far, Alexa can follow your voice commands and 0:07:57.120 --> 0:07:59.960 interact with the Internet as well as with other Internet 0:08:00.000 --> 0:08:04.040 connected devices. That list of Internet connected devices Alexa can 0:08:04.280 --> 0:08:07.280 work with is growing day by day, and Amazon is 0:08:07.280 --> 0:08:09.960 actually trying to build out the capabilities further and as 0:08:09.960 --> 0:08:12.440 such as hired a team to create a guide on 0:08:12.520 --> 0:08:16.320 how to develop for Alexa. I'm going to interview one 0:08:16.360 --> 0:08:19.040 of the developers on that team in a later episode. 0:08:19.480 --> 0:08:23.320 We actually have that scheduled for later this summer, and 0:08:23.320 --> 0:08:26.200 we'll talk more about what it's like to develop for 0:08:26.240 --> 0:08:29.760 this platform and the potential of using such a platform 0:08:29.840 --> 0:08:33.360 and in new and creative ways. So we have four 0:08:33.559 --> 0:08:37.000 really big players in the space. We've got Apple and 0:08:37.160 --> 0:08:42.440 Google and Microsoft and Amazon already vying to be the 0:08:42.480 --> 0:08:46.200 big digital assistant provider. Then we have the other names, 0:08:46.200 --> 0:08:49.560 like we've got the team behind Viv and other apps 0:08:49.600 --> 0:08:52.120 as well that are in this space that are trying 0:08:52.160 --> 0:08:55.280 to kind of become the voice that you interact with 0:08:56.000 --> 0:08:58.080 um so that it can do all the things you 0:08:58.120 --> 0:09:01.320 needed to do in a s seamless away as possible. 0:09:02.679 --> 0:09:05.400 So one of the things we need to also look 0:09:05.440 --> 0:09:08.320 at is how does this differentiate? How do these different players? 0:09:08.640 --> 0:09:12.000 How are they different from one another? If they're exactly 0:09:12.040 --> 0:09:14.440 the same as each other, then it really doesn't matter 0:09:14.480 --> 0:09:17.040 which one you pick, right, I mean, they're kind of 0:09:17.080 --> 0:09:19.640 depends just which platform you have available. If you have 0:09:19.720 --> 0:09:23.760 all iOS devices, then Siri is pretty much gonna be 0:09:23.800 --> 0:09:26.120 the one you're gonna depend upon the most most likely 0:09:26.120 --> 0:09:31.079 at any rate. So Cortana, Siri, and Google Assistant are 0:09:31.120 --> 0:09:36.079 all part of existing platforms like smartphones and computers, So 0:09:36.120 --> 0:09:43.040 they are incorporated into things that we already have. You know, 0:09:43.440 --> 0:09:46.560 you probably already have a smartphone or a computer or both, 0:09:47.240 --> 0:09:49.959 and so it makes sense that you would incorporate your 0:09:49.960 --> 0:09:52.840 digital assistant into that. You don't have to buy anything else. 0:09:53.040 --> 0:09:56.600 It's right there, and you can incorporate that into other 0:09:56.760 --> 0:09:59.560 systems that are connected to a personal network or a 0:09:59.600 --> 0:10:05.640 home net work. Then you've got Alexa, which debuted on 0:10:05.800 --> 0:10:09.839 a standalone device called the Echo, which again is just 0:10:09.880 --> 0:10:12.360 this sort of intelligent speaker, a smart speaker with a 0:10:12.360 --> 0:10:16.480 built in microphone. Google Assistant is actually following suit with 0:10:16.520 --> 0:10:19.480 that with Google Home that was announced at Google Io two. 0:10:20.280 --> 0:10:22.840 And Google Home is also a smart speaker with a 0:10:22.880 --> 0:10:26.440 microphone that's gonna be available sometime later in t and 0:10:26.520 --> 0:10:28.760 as of the recording of this podcast, I don't have 0:10:28.840 --> 0:10:32.200 a date or a price on that, so it's hard 0:10:32.240 --> 0:10:34.640 to say whether or not it will be competitively priced 0:10:34.640 --> 0:10:37.679 against the Amazon Echo. It does like it's going to 0:10:37.720 --> 0:10:42.760 be a particularly powerful version of this personal assistant, And 0:10:42.800 --> 0:10:45.640 there are also rumors emerging that Apple is also working 0:10:46.400 --> 0:10:50.040 on Sirie hardware, so it would be another standalone speaker 0:10:50.120 --> 0:10:55.320 microphone system of some sort, and that Apple's Sirie platform 0:10:55.400 --> 0:10:59.000 would exist on that. Now, as of the recording of 0:10:59.000 --> 0:11:01.320 this podcast, we don't have confirmation on that, so there's 0:11:01.320 --> 0:11:05.240 no timetable associated with such a thing or a price. 0:11:05.760 --> 0:11:08.920 I would expect that any announcement of such a device 0:11:08.960 --> 0:11:11.800 would come at one of Apple's big events, So probably, 0:11:12.520 --> 0:11:17.640 if I had the guess, I'd say September is when 0:11:17.640 --> 0:11:20.360 they would announce it. That's typically when they announced all 0:11:20.360 --> 0:11:25.320 the big iPhone changes. But that's just a guess. They 0:11:25.400 --> 0:11:28.319 might hold a single event for this particular thing, or 0:11:28.360 --> 0:11:29.760 they might not hold an event at all, They may 0:11:29.800 --> 0:11:33.040 just release it. That doesn't seem particularly apple like, but 0:11:33.080 --> 0:11:37.000 it's a possibility. So once the big deal with this 0:11:37.080 --> 0:11:40.160 technology in the first place, why should we care. Well, 0:11:40.200 --> 0:11:42.720 for one thing, it represents huge leaps forward in the 0:11:42.760 --> 0:11:46.440 field of artificial intelligence. So in one way, it's a 0:11:46.520 --> 0:11:51.320 really cool glimpse at the at the the state of 0:11:51.360 --> 0:11:55.600 the art in AI specifically, and stuff like speech recognition, 0:11:56.640 --> 0:11:59.040 which is pretty hard stuff. I mean, we all have 0:11:59.160 --> 0:12:04.319 different ways of pronouncing words, and depending upon your region, 0:12:04.960 --> 0:12:08.480 you might have an accent that has a different way 0:12:08.520 --> 0:12:11.960 of pronouncing word. For example, Uh, you know, the Brits 0:12:12.000 --> 0:12:15.160 say aluminium and we say aluminum here in the United States. 0:12:15.760 --> 0:12:18.760 Then even within a single country, you have different ways 0:12:18.760 --> 0:12:23.440 of pronouncing things. And when Google first began translating speech 0:12:23.440 --> 0:12:27.480 to text in voice messages, I noticed that it was 0:12:27.520 --> 0:12:32.120 having a real hard time interpreting the words of some 0:12:32.200 --> 0:12:34.920 of my friends and family. Now keep in mind I 0:12:34.960 --> 0:12:40.520 am in the Southeast United States Georgia, and that is 0:12:41.760 --> 0:12:43.840 we have a lot of people here with Southern accents. 0:12:44.600 --> 0:12:47.920 I have a tiny bit of one. My parents have 0:12:47.960 --> 0:12:52.559 a slightly stronger Southern accent. Some of my extended relatives 0:12:52.600 --> 0:12:56.360 haven't even stronger Southern accent, and so when they would 0:12:56.360 --> 0:12:59.920 call and leave a voicemail, Google had to guess at 0:13:00.040 --> 0:13:04.320 what they were saying, and was not always correct. I 0:13:04.320 --> 0:13:06.880 would have to go and listen to the voicemail because 0:13:06.920 --> 0:13:12.000 the transcription would be completely indecipherable. Now, over time this 0:13:12.120 --> 0:13:16.800 has improved. The speech recognition software has improved where it 0:13:16.880 --> 0:13:21.960 can adjust for things like different accents and and the 0:13:22.000 --> 0:13:25.000 different ways that people speak, using a lot of different 0:13:25.000 --> 0:13:28.800 algorithms that have been based in machine learning to kind 0:13:28.840 --> 0:13:31.920 of get a grip on what is being said and 0:13:31.960 --> 0:13:36.319 even and anticipate what the next thing will be in 0:13:36.559 --> 0:13:39.559 in any line of thought. Obviously, for someone like me 0:13:40.040 --> 0:13:44.040 who stumbles over words occasionally, that's a real challenge because 0:13:44.080 --> 0:13:46.120 sometimes I don't even know what's next to come out 0:13:46.160 --> 0:13:48.960 of my mouth. But that's really where where that power 0:13:49.040 --> 0:13:53.640 comes in. Now, over time, not just speech recognition has improved, 0:13:54.080 --> 0:13:58.360 we've also had to look at the problem of natural language. Now, 0:13:58.440 --> 0:14:00.720 natural language is how you and I could communicate with 0:14:00.760 --> 0:14:04.400 one another. Unless it's like a really formal setting. We 0:14:04.520 --> 0:14:07.360 usually are pretty casual with our language, and we can 0:14:07.360 --> 0:14:10.960 make use of lots of different linguistic flourishes and tools, 0:14:11.040 --> 0:14:16.040 things like figures of speech, metaphors, similes, puns, references, and 0:14:16.080 --> 0:14:19.040 lots of other stuff that gives meaning to what we say. 0:14:19.080 --> 0:14:22.480 But only if the other person also understands what's going on. 0:14:23.160 --> 0:14:26.960 They also have to have that benefit. Otherwise it just 0:14:27.040 --> 0:14:30.520 becomes a jumble of nonsense. I'm reminded of a Star 0:14:30.600 --> 0:14:34.840 Trek the Next Generation episode where characters only spoke in 0:14:35.240 --> 0:14:40.080 um allegory, and if you didn't have that cultural background, 0:14:40.160 --> 0:14:43.560 if you didn't understand the references, you didn't understand where 0:14:43.600 --> 0:14:49.200 the community, what, what the communication meant um. Similar problem 0:14:49.240 --> 0:14:52.760 with machines. They don't necessarily know what we're saying all 0:14:52.800 --> 0:14:54.800 the time. A lot of machines are not very good 0:14:54.800 --> 0:14:58.560 at doing this. But natural language familiarity has been a 0:14:58.640 --> 0:15:02.640 huge challenging in AI, and we're getting better at overcoming 0:15:02.680 --> 0:15:05.400 that challenge. So at that that same Iowa event where 0:15:05.440 --> 0:15:09.560 Google announced Google Home, they demonstrated that you could start 0:15:09.560 --> 0:15:13.400 a conversation with your personal assistant asking something fairly specific, 0:15:13.480 --> 0:15:17.280 such as, we're gonna go with a local reference for 0:15:17.280 --> 0:15:22.200 for yours truly, how are the Atlanta Braves doing this season? Then? 0:15:22.240 --> 0:15:24.720 The assistant would actually break your heart by telling you 0:15:24.760 --> 0:15:27.280 how poorly the Braves are doing this season and it 0:15:27.400 --> 0:15:31.160 is abysmal. And you could follow that up with when 0:15:31.200 --> 0:15:34.280 do they play at home next? And the assistant would 0:15:34.400 --> 0:15:38.160 understand that when you say they, you mean the Atlanta Braves, 0:15:38.320 --> 0:15:40.960 and when you say at home, it would understand you 0:15:41.000 --> 0:15:45.120 meant Atlanta, Georgia. So it would be able to figure 0:15:45.160 --> 0:15:47.720 out the context of what you said without you having 0:15:47.760 --> 0:15:52.320 to restate when do the Atlanta Braves play in Atlanta next? 0:15:53.080 --> 0:15:55.840 You could take these little linguistic shortcuts that we would 0:15:55.840 --> 0:16:00.640 normally do in natural conversation, but typically machines are not 0:16:00.760 --> 0:16:04.720 great at that. They don't have the capacity to understand 0:16:05.320 --> 0:16:10.120 how one sentence can follow another. But this is an 0:16:10.160 --> 0:16:13.600 example of how that's changed through machine learning. So you've 0:16:13.600 --> 0:16:17.720 gotten this new approach where you can continue a series 0:16:17.760 --> 0:16:22.000 of questions that build on previous questions and answers, and 0:16:22.080 --> 0:16:26.119 the Google Assistant can continue to give you relevant information, 0:16:26.200 --> 0:16:31.000 which is a pretty powerful statement in AI. Also, you 0:16:31.080 --> 0:16:34.160 might have heard that funny story that Google fed romance 0:16:34.280 --> 0:16:37.080 novels to its AI to make it better at understanding 0:16:37.160 --> 0:16:40.880 natural language, And to be fair, that's just part of 0:16:40.880 --> 0:16:43.800 the story. Google actually fed lots of different types of 0:16:43.920 --> 0:16:47.720 unpublished literature to its AI, all with the goal of 0:16:47.800 --> 0:16:50.560 teaching the AI that there are many different ways to 0:16:50.640 --> 0:16:53.760 say the same thing. So here's an example. Um, I 0:16:53.760 --> 0:16:57.080 could say it's raining pretty hard today, or it's really 0:16:57.120 --> 0:17:00.280 coming down out there, or it's raining cats and dogs dogs, 0:17:00.360 --> 0:17:03.400 or it's pouring outside, and all of those mean the 0:17:03.440 --> 0:17:05.920 same thing. But they're all different ways of saying that 0:17:06.000 --> 0:17:11.080 it's raining really hard. And there are a lot of 0:17:11.080 --> 0:17:13.919 other ways I could say the same you know, to 0:17:13.920 --> 0:17:17.360 to express the same thought using different words. And that's 0:17:17.359 --> 0:17:22.720 a challenge for machines because we as humans understand that 0:17:22.840 --> 0:17:24.640 you can say all these different things and that all 0:17:24.760 --> 0:17:28.159 means the same thing. But machines have to be taught that. 0:17:29.800 --> 0:17:32.320 So romance novels, as it turns out, are a good 0:17:32.359 --> 0:17:35.639 way to teach a an AI how to interpret different 0:17:35.640 --> 0:17:41.120 things because romance novels are incredibly formulaic. If you were 0:17:41.160 --> 0:17:44.879 to break down a romance novel and and you outlined 0:17:44.920 --> 0:17:48.199 it scene by scene so that you understood where the 0:17:48.240 --> 0:17:51.040 beats and the story were and who the characters were 0:17:51.040 --> 0:17:53.720 in their relationships to one another, you would see that 0:17:53.800 --> 0:17:57.119 a lot of romance novels follow the exact same structure, 0:17:57.320 --> 0:18:00.959 the exact same plot structure, but because they're written by 0:18:00.960 --> 0:18:05.120 different people, because the character names in places are often 0:18:05.200 --> 0:18:07.520 changed from book to book. I mean, obviously you wouldn't 0:18:07.560 --> 0:18:11.200 want to write the same novel forty times. It means 0:18:11.200 --> 0:18:13.840 that you have a lot of different ways to express 0:18:14.040 --> 0:18:18.119 the same ideas. So if you feed a whole bunch 0:18:18.119 --> 0:18:22.520 of formulating novels into an AI to teach it humans 0:18:22.560 --> 0:18:25.640 have lots of different ways to express the same thoughts, 0:18:26.359 --> 0:18:29.880 that's a pretty powerful tool. Um. And again, it wasn't 0:18:29.920 --> 0:18:32.480 the only type of story that was being fed to 0:18:32.960 --> 0:18:35.680 Google's AI. It's just the one that caught a lot 0:18:35.680 --> 0:18:39.359 of people's attention because it the headlines right themselves at 0:18:39.400 --> 0:18:42.679 that point. So one thing that is really, you know, 0:18:42.760 --> 0:18:44.720 funny about that is a lot of people made jokes 0:18:44.720 --> 0:18:48.439 about Google AI suggesting different ways to rip a bodice 0:18:48.600 --> 0:18:52.040 or to make a bosom heave from the whole romance 0:18:52.080 --> 0:18:54.080 novel thing, But as it turns out, it was there 0:18:54.119 --> 0:18:58.800 was some real thought given to using this approach. Now, 0:18:58.880 --> 0:19:01.959 one of the ways that these assistants work so well 0:19:02.040 --> 0:19:05.359 is to tap into information about you and to store 0:19:05.400 --> 0:19:08.560 all of that off of the hardware so that it 0:19:08.640 --> 0:19:12.320 can anticipate what you want and what you need and 0:19:12.359 --> 0:19:16.240 how to fulfill that. So, for example, if I'm using 0:19:16.280 --> 0:19:20.720 Amazon and I'm using the Echo and I'm I'm using 0:19:20.880 --> 0:19:25.680 Alexa to purchase certain things off the Amazon Store, this 0:19:25.760 --> 0:19:30.840 ends up tapping into that algorithm that tells Amazon what 0:19:31.040 --> 0:19:34.600 I've bought and what I have browsed and and the 0:19:34.600 --> 0:19:36.680 sort of stuff I'm interested in, so it can suggest 0:19:36.760 --> 0:19:38.920 new things that I might be interested in but didn't 0:19:38.960 --> 0:19:41.680 know about. All of that is a very powerful tool. 0:19:42.760 --> 0:19:48.520 One of the exceptions here is Apple's SIRIE. So Apple 0:19:48.600 --> 0:19:52.240 pretty much locks everything down into the hardware as opposed 0:19:52.240 --> 0:19:55.159 to sharing it with third party or putting it in 0:19:55.200 --> 0:20:01.320 the cloud. That's because Apple's revenue source selling that hardware 0:20:01.440 --> 0:20:04.959 and related services like support plans like like a product 0:20:05.000 --> 0:20:09.800 support or protection plans for your hardware. That's how Apple 0:20:09.840 --> 0:20:13.040 makes its money. It's making it through selling this hardware 0:20:13.080 --> 0:20:16.400 that it is producing, as opposed to something like Google, 0:20:16.480 --> 0:20:20.720 which until Google Home comes out, it's selling an idea 0:20:21.240 --> 0:20:25.640 to you and then selling you to advertisers. Uh. So 0:20:26.800 --> 0:20:29.320 Apple That the benefits Apple in some ways because it 0:20:29.359 --> 0:20:32.200 means that you can trust Syria a little more than 0:20:32.240 --> 0:20:35.480 you could some of the other assistants because it's it's 0:20:35.520 --> 0:20:39.280 mostly contained to your device. On the flip side, it 0:20:39.280 --> 0:20:42.359 makes the actual service a little less useful because it 0:20:42.400 --> 0:20:45.520 cannot tap into the massive resources of the Internet the 0:20:45.560 --> 0:20:50.400 way some of these other assistants can, because again it's 0:20:50.440 --> 0:20:53.320 all pretty much contained your device. Now I can access 0:20:53.320 --> 0:20:55.760 I can pull stuff from the Internet for you, but 0:20:55.800 --> 0:20:58.960 it's not as interactive as some of these other assistants are. 0:21:00.480 --> 0:21:05.080 Uh So, there with the possibility of advertising or things 0:21:05.160 --> 0:21:11.040 like Google or Amazon rather Amazon's integrated shopping services, you 0:21:11.080 --> 0:21:15.040 start to see some real potential for revenue generation on 0:21:15.080 --> 0:21:17.879 the back end. But it also brings up some questions 0:21:18.359 --> 0:21:22.840 about privacy and security. Now. To look into that matter further, 0:21:23.040 --> 0:21:26.440 I spoke with an expert on the subject, the founder 0:21:26.520 --> 0:21:30.159 of a company called Big I d Dmitri Serota, and 0:21:30.200 --> 0:21:33.800 here's what he had to say. Well, I think that 0:21:34.000 --> 0:21:37.679 clearly there's a certain degree of inevitability around this. I 0:21:37.720 --> 0:21:41.280 think we've moved from an age of having these technologies 0:21:41.320 --> 0:21:42.800 and you can almost think of this as kind of 0:21:42.800 --> 0:21:46.679 web dot dot one in terms of being responsive to 0:21:46.760 --> 0:21:50.879 the user and personalization really being about kind of targeting you. 0:21:51.240 --> 0:21:54.639 I think we're now shifting to an era of anticipation. 0:21:55.119 --> 0:21:58.320 You know, the technologies are becoming smarter and they know 0:21:58.480 --> 0:22:01.879 more about you because they uch you on so many levels, 0:22:01.880 --> 0:22:04.280 whether you're on the web, whether you're on your mobile, 0:22:04.359 --> 0:22:06.640 whether you're at the office, whether you're in the car, 0:22:07.240 --> 0:22:10.399 whether you're at home, as is the case with with 0:22:10.480 --> 0:22:15.000 Google Home and Amazon Echo and similar technologies, that they're 0:22:15.040 --> 0:22:19.760 no longer just about kind of responding to a particular action. 0:22:19.920 --> 0:22:23.239 They're now trying to anticipate what you'd want. And in 0:22:23.280 --> 0:22:25.600 some degree, you know, they're becoming more like your mother 0:22:25.720 --> 0:22:29.239 or your parents, where they know so much about you 0:22:29.359 --> 0:22:32.600 that they anticipate your needs um and there's a good 0:22:32.640 --> 0:22:36.240 and a bad to that, right, So just the actions 0:22:36.280 --> 0:22:38.840 that we would take in our homes can start to 0:22:38.880 --> 0:22:42.720 set up these these expectations, for lack of a better word, 0:22:42.720 --> 0:22:46.280 that our technology will have about us. For example, the 0:22:46.800 --> 0:22:50.080 easiest way I think to illustrate this to today's audience 0:22:50.119 --> 0:22:52.720 is to talk about something like the nest thermostat, where 0:22:53.240 --> 0:22:55.720 you have said it a certain way, and it starts 0:22:55.760 --> 0:22:58.520 to learn what your preferences are over time and then 0:22:58.560 --> 0:23:01.439 it begins to automatically just without you ever having to 0:23:01.440 --> 0:23:04.720 touch it, to the point where it's even seeing quote 0:23:04.800 --> 0:23:07.800 unquote seeing when you are home versus not home. This 0:23:07.880 --> 0:23:10.199 is the sort of stuff that, when incorporated into a 0:23:10.240 --> 0:23:13.320 device like Google Home, can become very powerful. But also, 0:23:14.000 --> 0:23:17.199 like you said, has this this other side to it, 0:23:17.240 --> 0:23:20.680 this side that that if we don't pay attention to it, 0:23:20.680 --> 0:23:25.840 it could become potentially um harmful to us, or at 0:23:25.920 --> 0:23:30.040 least at the at the very least anyway uh inconvenient 0:23:30.160 --> 0:23:34.480 to us. So, for example, with our setting up Google 0:23:34.520 --> 0:23:37.360 Home so that we would be able to control lighting 0:23:37.480 --> 0:23:41.640 and security systems and uh thermostats, not only would we 0:23:42.080 --> 0:23:44.439 have it set up so that it's to our preferences, 0:23:44.440 --> 0:23:47.399 but it actually has learned when we're at home versus 0:23:47.400 --> 0:23:50.520 when we're not at home, and what that information means 0:23:50.600 --> 0:23:55.280 could be potentially very harmful to us. So in your mind, 0:23:56.280 --> 0:24:00.359 where where does accountability lie? Is this something that we 0:24:00.480 --> 0:24:03.640 ourselves are are at least partly accountable for that kind 0:24:03.680 --> 0:24:07.360 of information? Are the companies that create this technology? Are 0:24:07.359 --> 0:24:11.119 they accountable? It's it's such a cloudy area. Where do 0:24:11.200 --> 0:24:15.439 you see that? So it's a mix now clearly, and 0:24:15.480 --> 0:24:17.520 I think I want to kind of emphasize this. You 0:24:17.560 --> 0:24:21.880 mentioned earlier how this could become inconvenient. The reality is 0:24:21.880 --> 0:24:24.840 is that we as the consumer want this because we 0:24:24.920 --> 0:24:28.840 want it because it is convenient. We want technologies that 0:24:28.880 --> 0:24:32.280