Native Voice CEO John Goscha on the Evolution of AI Voice Assistants

foreign speaks we're talking about artificial intelligence and AI Voice assistance how the growth in AI is driving greater choices and more sophistication in the market for voice assistance it's not just about Siri and Alexa anymore to discuss that I'm joined by John Gosha founder and CEO at Native voice John very good to have you with us today James great to be here thanks for having me on so obviously there's a lot going on with AI this year but ever since chat GPT debuted at the end of last year it's been almost kind of a semi Asteria uh what do you see as AI is growing so quickly this year what are a couple of key trends you see driving in the market yeah it's pretty unbelievable every day you have to wake up and spend an hour or two just catching up on what's happened right certainly is moving quickly but you know the piece that I find most interesting is you know now we can talk to a data set whether that be a document a book a set of documents a database and so that unstructured data we can now have a conversation with and so I think about where all those data uh you know depositories lie and you know a corporation has a lot of you know documents and information that might be in different places and so uh you know that's kind of an interesting area all of us have our own data right we have emails and text messages and and documents um and so I think that being able to talk to a specific set of data say you want to ask a question of a set of reviews when you're on Amazon or you want to ask us a question of a product of a website a particular company so I think there's some really interesting consumer use cases but also for the Enterprise and so uh you know every company I know is now talking about and thinking about you know how do we allow our internal knowledge uh to be more accessible by our employees but also how do we allow our products and services to be more accessible to our customers and so I follow you know where's the data and I don't mean just big databases but you know documents and text and you know now we can interact with those and ask questions in a very unstructured way and so those are the pieces that you know I see you know heating up first there's a lot more to come but that's the you know the area that's getting a lot of attention and and that we paid attention to being able to talk to to a system actually use your voice and speak with it does change our relationship with computers so so dramatically I think about it's really sort of like the democratization of of AI in that it used to be just accessible only a few data scientists or real experts and now every single person can now talk to their system in one way or another yeah it's it's pretty incredible the uh the way it changed the interaction you know the example that you know I think about is you know I might read a book and others will read that same book and you know you'll have a couple key takeaways or when you're reading an article you'll skim down to just the pieces you want so consuming information you know there's a lot of things that we don't take away and so there's a lot of noise and I compare that to you know having a conversation with someone you can ask a question and you can drill down and ask more questions about that or come up and ask about a different topic and drill down and so that method of being able to get what you want find out the information is now possible with you know books and other texts and so now you don't have to consume and go through all the noise to get to the piece that you want and so I think it'll fundamentally change you know how we learn how we you know consume knowledge and that's that's pretty interesting hey fascinating really so I I know you're really close to the AI Voice Assistant market so how do you say ai's role with voice assistance evolving these days so I think it's going to be transformative and you know I I hear a lot that oh you know the voice assistants aren't that smart they they don't do a great job um you know we all have that frustrating moment where you're asking Alexa or Siri for something and they respond you know I don't understand or I can't do that right and so yeah I think that will fundamentally change I was speaking to one of my advisors Larry heck who ran speech at Microsoft and became CEO of Samsung Bixby he's now at Georgia Tech and he said John I remember walking around the campus of Microsoft with Adam Shire the founder of Siri and this is before Apple bought them and they were talking about how the big innovation 10 years ago was speech to text we could take what someone was saying and get it accurately in the text at a rate 95 or greater and so that gave us Alexa that gave a Siri that gave us Google that Innovation what's happened today is now with large language models we can Now understand and what the person means what's the essence of what they're truly asking for and so today you have to learn how to talk to a voice assistant to learn the commands it can do but now with large language models you know you'll be able to speak much more freely and it'll understand what you want and so I think that there will be less of a structure kind of categorization system and something that's much more broad that you can um you know have a much more fluid conversation with so that's the first piece that I think we'll experience in Voice assistance that will change let me ask you one about a question about that is that because the large language models are being trained by us with all of all of our various speech patterns like what is making them more versatile yeah so I think maybe a good comparison is we've all gone to uh you know chat gbt or you know one of the other uh large language models and we put a piece of text in and we ask it to summarize it so it's very good at summarizing information and you can even say you go give me it in two words and I'll grab the best two words that describes you know a whole page of text so we've all have experienced that but what it's doing is it's able to you know find the words that you are getting at the intention the essence of what the person has said and so I'll give you an example where a small investor of ours is Chick-fil-A and uh we have some recordings from the drive through and you know if you think you'll ever listen to yourself or someone else ordering you know Chick-fil-A you know it's oh I want a spicy chicken sandwich actually no pickles oh actually make that a combo this this and that and it's all over the place right a year ago we couldn't ingest that we couldn't understand what the person wanted today with a large language model we can ask it and say okay what does this person really want and it gives us the exact order of they want spicy chicken sandwich no pickles this modification this and that and now we can ingest that into a two-layer system where we have kind of your traditional Voice Assistant with a large language model on the front end understanding what you actually wanted and so it's uh you're very good at summarization which is you know very you know similar to getting at the the intent of what someone wants all right so that that's that's the summarization what else about how ai's role with voice assistance is evolving so you know one uh yeah there's a front end of course with a large language model um but also the back end you'll be able to respond to you more naturally so those things are great but what about kind of the the meet in between and so what we see there is using Vector sensors uh you know Vector databases um you know I guess I could think of these as kind of small language models um where you can take a set of data and you can talk to it like we were discussing earlier but you don't need to train that model you don't need to train that data and so I'll give you an example we have been working with a a large audio library the largest library of audio content in the world and they have 92 000 stations and you you may ask this Library oh I want to hear you know something that is similar to Pink Floyd and Led Zeppelin and some small little town in um you know in the Pacific Northwest and you know it'll know that okay well those artists are you know rock artists because on the internet their names appear with the word Rock many times and it'll also know that that little town you might have mentioned is near Seattle and not because it knows geography it's not you know according Google Maps it just knows that Seattle appears close to that word and and you know on the internet and I'll come back with you know here's a rock station in Seattle you might like and so we never had to train it to know where Seattle was or to train it that those artists are rock artists it's able to understand that because it has the large language model um you know in the background running and so it'll be able to give you better responses um based on you know descriptions that are maybe less specific to Native voice how is native voice serving The Voice Assistant needs of its customers what what is a native voice Advantage John yeah so we talk about our customer I mean our customer first and foremost is the end user and you know we want to help the end user get more done throughout their day simply by using their voice and you know I think it's interesting that you and I we use our eyes our ears and our mouth to communicate with one another you know even here some of the um communication is body language most of its audio sure but I think it's strange that we've been interacting with computers using our fingers for decades with mice and keyboards and and touch screens and so you know there are certain things that the consumer wants faster easier so you can go from having something on the tip of your tongue to having it in your ears if it's a piece of audio content or requesting a product or a service and so that's where we're focused is on that that end user but this is really about building an ecosystem where I have confidence that Uber will build the best experience for calling a car and Starbucks for coffee Etc and so you know those Brands when they focus in their area of expertise I'm learning they do a better job so they bring a better experience to the end user but also for the end user we all have brands that we know and trust um you know you go to Uber for a car you don't Google a car for example sure and so when you have more guidance as to what you can do with a particular brand it makes it easier to interact with a voice assistant you don't have that white sheet of paper problem like we had with the general voice assistance where you say well what can it do you know I just don't know and so that's where you know native voice we're allowing Brands to bring these incredible conversational AIS to Market and connecting them on our network of millions of devices so they're available for you and I as we're you know walking down the street with our earbuds in or driving in our car voice interact with say Siri or or Alexa if at all yeah uh we have the major assistance on our platform uh of course uh Alexa uh Google and Siri and so uh we give you choice so you can talk to whichever assistant you might might prefer or might do the best job um but then we also bring multiple conversational AIS or voice assistants alongside them and so you were able to have the greatest amount of choice but you know talk to Alexa if you want to find out where your package is but you can also ask I heart to you know play your favorite artist or your favorite station and so uh we give you the ability to you know choose which voice assistant or conversational AI is best for that particular job that particular task so someone is actually looking at a menu of choices and they might choose a Siri or Alexa or iHeart correct so yeah quick user story is so one of our uh great customers is Skullcandy headphones the second largest uh or second best-selling headphone in the country and so if you buy a pair of Skull IQ headphones from Skullcandy pop them in your ears you download the app and the app you can simply turn on and say Yes I want to have Alexa I want to have I heart I want to have you know a hey Skullcandy Etc and so each time you want to go on a run you leave your phone in your pocket put your earbuds in your ears and you can say hey I Heart play a running playlist and as you're running do you see clouds rolling in you can say you know hey Alexa you know is it going to rain today and all in this seamless experience with uh your head up don't have your phone you know in front of you uh you know burying um yourself in that screen time and so um you know that experience is is what we enable for the the end user a snake so I think the big picture is is the future of AI and The Voice Assistant I know it's going to get extremely capable in the years ahead uh what what are some key Milestones you think we can see in the next oh say half decade or or so if you look that that far out yeah so you know the near term we talked about just being able to get what you want faster um be able to ask a question and avoid all the noise get the answer dig deeper and so um you know I think that certainly is happening today I do think that every application every app every website will be rebuilt in a way that is custom to you it doesn't have to be a structured where today if you go to buy a product you look at the product and then you go and need to put in your shipping information and then your credit card and there's this kind of process and tree well conversationally you don't need to have all that structure um you can say you know yeah I want to you know now put in my credit card number or I want to do this piece and actually kind of go in any order that you'd like and so I think that we will fundamentally change how we interact with our products our services you know our apps and our websites and so I think that they will be designed custom to us and we can interact with them in a way that's you know very flexible and and customized and so I think that will uh you know certainly be a second big trend a third one and the last one I'll mention that is really interesting to me is our personal data um you know if you uh think about all your emails all your text messages you know all your documents I'm a big advocate of personal ownership and control of our data I think that'll be one of our greatest assets into the future but uh you know with all that information you know we can learn a lot about ourselves um we can you know get insights on different decisions we can share access to uh you know others on and our knowledge and so if you think about someone you know well you know maybe as a parent maybe it's a sibling you know ask themselves you know ask yourself a question you're ahead of that person right now and you can hear that person respond in a way they would probably respond and their tone you can actually hear them right well the reason for that is you have all this experience with that person and now you can you know pattern match in your head and have a pretty good idea what they're going to say but now that we have conversational AI and we have you know large language models our personal data will be able to access our own knowledge and other people's knowledge in ways that we never were able to do before you know usually when we all pass away we you know take a lot of our experience a lot of our knowledge with us unless we wrote a paper wrote a book or have an apprentice and so I think with all of our data now knowledge will be carried forward in a way that has never been possible uh for Humanity but also all that information on us we'll be able to get better products better Services because we understand ourselves much better and with permission the companies that are serving us can understand us as well and so that's something I look at you know five years or longer from now that will be uh a real change in the way we interact with knowledge and and customization it's the interesting idea it's almost sci-fi it's almost like we're going to be talked to a a recording of you know grandpa so to speak and he and his voice assistant we continue to talk with us even though he himself is no longer with us is that what you mean personal day you you could do that I think that's uh definitely possible but um you know think about you know an expert that you don't have access to uh that you want to ask a question to and they don't have time to answer everyone's question right um so you know we're starting to break the boundaries of of time and be able to share knowledge um and access word we only are able to talk to people who are within our our reach um but if you know someone has all their you know all their personal information their personal AI you know we can learn from them uh we can get advice or you can ask you know not just one person ask a thousand people that have something in common and see what what that what they say so I think it'll fundamentally change how we share knowledge um and how we retain knowledge that is really interesting John I think you said it a lot of good stuff it's going to be a fascinating sector to watch in the years ahead I hope you come back and talk with us again sometime nice all thank you for having me foreign

This transcript was generated automatically from the video's captions and may contain errors.