How Generative AI Is Changing Travel — With Priceline's Marty Brodbeck
Channel: Alex Kantrowitz
Published at: 2023-12-12
YouTube video id: HX_YuEbhE4g
Source: https://www.youtube.com/watch?v=HX_YuEbhE4g
hello YouTube we're here with a special edition of our conversations for you sponsored by Priceline we're going to be talking about how generative AI can change the travel space we've heard so much about Theory and we're trying to get into practice we've talked about it in the terms of public relations and marketing now let's talk about how it can change travel joining us today is a very special guest Marty broadbeck is here he is the chief technology officer of Price Line Marty welcome thanks for me thanks for being here let's go back to last November chat GPT comes out do you immediately think hey this could revolutionize the way that we do business or is there a bit of time that it takes to set in before you think about maybe we can use generative AI in the way we reach customers the way that we position ourselves the way we do business internally well when when chat gbt came out we we thought that this potentially is a ground piecing ground piece of technology that could fundamentally change how customers interact from a travel perspective we as a company have been have been using machine learning for a long time and know the power of what it can actually achieve and accomplish from an Automation and a customer perspective and so when we saw the the launch of chat EBT we immediately started brainstorming on ideas and areas of the company that we could be applying this kind of capability towards and where has it gone into practice for you because you know I think there's there's this question of do people actually want to discuss their travel plans with the chatbot or would they just prefer the traditional graphical user interface that you have on a priceline.com right now yeah so we we immediately started looking at really four areas where we felt generative AI could be really applicable to uh our company and our products and more importantly to to our customers um the the first base was really embedding generative AI capabilities into our products such as our checkout flow and our post booking flow uh the second is we thought that a generative AI capability like Chad gbt or Bard or cohere could have real relevance in our post booking process in the way in which customers interacted with our company post booking um and had questions around products fairs uh fees Etc as it pertains to the products that they bought the third is really around how we generated marketing content and how we apply the technology to provide richer content to our customers about destinations or cities that they were potentially traveling towards and then the last was we took a hard look internally about how we could apply generative AI capabilities to further Drive greater levels of productivity within the company particularly within you know my organization which is product engineering great and so I want to talk let's go through one by one as we through this conversation and talk about them first of all in terms of like the actual consumer experience interacting with price line going to go back to that question before people are there's a debate right now really in terms of whether people want to stick with the traditional website architecture to book their travel and ask questions about or or or find information about it and how useful something like a chatbot might be in terms of booking travel finding destinations you know maybe people are thinking I can do my entire itinerary just by saying to a chapot hey you know find me some Island in gree that I can stay at for seven days for you know $2,000 a person I'm just throwing numbers out there so I'm curious from like your perspective you're the one that's going to you know help enable this where does your thinking land on that front well we're a very data driven company so we started embedding G of AI capabilities right into our checkout flow where if a customer had a question around a hotel a flight a product that before they booked that they could have the experience of asking questions about the product that they were potentially booking and so we we Dove right into the heart of our customer Journey went right to the checkout flow where Not only would we enable uh customers to ask questions and get answers about the products that they were about to book but also embedded the checkout flow right within the experience of generative AI to allow them to have a frictionless experience that once all their questions were answered that they could immediately just go to purchase uh so that was where we started and more importantly because we're highly AB D AB testing driven we ran tests to make sure that the bot had a positive impact on the user experience in the you know in revenue and after several iterations of prompt engineering we got to a point where the bot was AB testing positive and so we pushed it out to 100% of our our customers on our platform the next part that we tackled which we recently just launched is putting generative AI into the post post booking process to answer any questions that customers may have around any of the products that they were booking after their purchase uh and again because we're a very AB driven tested company went through several iterations of you know prompt design and iterations and got to a point where uh in our post booking process the bot had a positive impact on the consumer experience now we shifted our Focus towards what we're calling the travel concierge part of our business which is really the trip planning and we're currently working on capabilities to allow users to do free tests free Tech searching so I'm planning to be in San Francisco tomorrow at 10:00 am. can you recommend a flight and a hotel capability or a set of products that would be relevant based on the meeting that I'm going to have next day and then walking the user through the recommendations and the ability to do trip planning as well um more importantly I think you know it's changing the Paradigm for how people search products um rather than having the your typical Search bot uh search box where you have to put in when I'm when I'm going on the trip and when I'm coming back having it be more of a conversational experience using natural language rather than you know strict text boxes and that's what we're currently developing another thing that we're super excited about is embedding generative AI capabilities provide a richer experience for our customers when booking hotels there there are several things that customers are very interested in number one points of interest around the hotel uh or the area that their hotel that they're potentially booking is located to the neighborhood description of where the hotel is particularly in major cities like New York Los Angeles Las Vegas London and Paris the neighborhood description is key in making the decision about not only what can I do uh on this trip but what potentially my family can do uh so that's very important hotel reviews so summarizing hotel reviews that um that people have put onto our website is very important for people to make a decision and that's really using generative AI to basically provide these capabilities embedded in our e-commerce platform as well so that those to me are game changers in terms of providing a much richer personalized experience for our customers rather than a two-dimensional one that's based on when I'm going on the trip and when I'm coming back on the trip yeah definitely I mean I'm thinking just about searching for hotels and what I typically do is search for hotels first I go location then I go uh rating and then I look if it has good Wi-Fi and I disqualify it at the end if the Wi-Fi isn't good yeah and I imagine with a chatbot you could almost get that all together being like can you find me a hotel in like Union Square in San Francisco it has to be above an 8.0 rating and the Wi-Fi has to be good and it can comb through those responses and say hey here's some that people generally like the Wi-Fi so it sort of saves a tremendous amount of time one of the interesting insights we got from our AB testing is a lot of our customers had uh questions about hotels being pet friendly really uh and so being able to provide answers to those questions in real time for our customers as they're planning their their trip has been you know super helpful for us you know Marty one of the things that I T to tend to find when I'm searching on something like Bing or Bard is um it can find information that you wouldn't generally find in the first like 10 results on search right it surprises you with information and there's only so much information you can hold on like the front page of a of a travel website like price line so does does um the the search functionalities that you've that you're enabling with this bot does it offer surprising or like deep deep down in the data results that people can you know end up making better choices with when they encounter it well one of the one of the great things around generative AI is that you can marry it with your own data sets internally which uh for us has been gamechanging so what what we're doing within our platform is we're taking all of our hotel content and putting it through a machine learning pipe line and Maring that data our own data with authoritative sources of location information and map information to provide complete detailed accuracy on things such as hotel reviews neighborhood descriptions points of interest uh the second advantage to taking this approach it also reduces the amount of hallucinations that you would get from a large language model so marrying the large language model with our own data sets with thirdparty data really drives down uh the amount of hallucinations that we get from these kinds of conversation and almost creates authoritative sources of cach information around hotel reviews points of interest neighborhood descriptions um that allow the site to perform faster because the other thing that to to take into account with all these things is the cost um cost in and running and managing a generative AI platform could Skyrocket so you have to be very smart about when you refresh or make direct calls to the large language model and this third-party information yeah and how about knowing the customer so I think that one of the things that everybody's looking for when it comes to these Bots and a real point of frustration using something like a chat GPT or a Bing is it has like Goldfish Memory right so you go say hi you have a long discussion you talk about your life with it and it has you know it's very empathetic conversation with you you come back 10 minutes later totally forgets who you are now when it comes to travel memory seems like it's something that could be so huge because if price line for instance can remember what type of hotels I like or that Wi-Fi is important for me and I go into that bot and it says Hey welcome back okay we we see that you're interested in traveling to Austin for instance and and given your past you know there is a little bit of this in the product products right now you know but but it would be amazing to say like given your past uh uh bookings we know you like places with good Wi-Fi generally downtown close to a good coffee shop you know ratings above 8.0 nice to see you again here's what we found um as opposed to me having to reintroduce myself over again so where are we in the state of like being able to get to that point well I think this is more about how companies design and develop their data AR architectures then generative AI understanding personalization on the individual level so over the last several years we've made huge investments in building out a customer data platform that actually understands who you are what your booking preferences are what you've searched what you've booked when you've booked have you booked as an individual traveler or have you booked as traveling with a family and what have been the nuances of hotels and flights you've taken as an individual versus as a family and so what we've done is we've married all that rich CDP data so that when you log into the generative AI experience a signal was fired that understands who you are so in this case you're Alex and I understand what you searched on what you've booked in the past what are your preferences so that as we're going through the conversation it completely understands the background that you have as a customer based on all the intelligence we built into our data infrastructure so that way it's a very personalized experience um in that conversation to understand all those preferences and that could just play right into the bot correct very cool so one of the things that I've been wondering now that we've been talking a little bit is you mentioned that the first place that you looked was in the checkout process how can we you know deploy this there and you know for me it it seems interesting because intuitive like like going by my gut I would say maybe at the beginning when you're in Discovery or at the end when you have customer service issues but you're saying at the checkout so what kind of questions come out at checkout and why start there um we felt that the checkout was really the folr point where customers have the most questions about a booking like they've already are knee deep into the details page of one of our products so say it's a hotel and they're about to purchase something but they have some you know just nagging questions that could be plaguing them like a great one is is it pet friendly um what are the points of interest close to this hotel uh what's the neighborhood description around it so you know answering those final closing details for the for the consumer builds more confidence in their booking and that's why we we started with with checkout for for from our perspective two it's at the heart of what we do uh so creating a frictionless checkout experience that answers all questions for our customers before their booking we felt was a great place to start very cool okay let's talk about marketing uh you mentioned that this has been used to basically create high quality content at scale for Price Line uh I can't even imagine the scope of content that you're working to create and so that's why I want to ask about it um how does this help and where what type of content is is it creating um how are you again preventing from hallucinations from taking hold and how are people responding to some of the content that you're using generative AI tools to put out there so the the marketing content is all about creating merchandized Pages uh blogs um detailed ad Snippets you know profiling a descriptive um city for instance so we we've we've taken the top 1,000 City destinations in the world and have been systematically creating content in those three different channels I I talked about you know a part of a large part of our our business is is indirect business that comes through people searching so providing very high quality merchandise personalized content around these cities we felt was a great way to apply generative AI uh on the second on the second part of your question around hallucinations uh we've built a hallucination service that verifies the information using third-party information sources like um in the case for a lot of our our hotel products we're using places API from Google um which is really an authoritative source for a whole host of geographical uh information and so as part part of our pipeline there's a hallucination check that occurs to ensure the accuracy of the information that we're given we also have another check in that process which is really human copy review where we have humans review the copy uh being generated after our hallucination check just to make sure that there's a level of accuracy for the information that's going out yeah that monitoring it's sort of like how people's jobs change it instead of like actually creating the product you monitor the inputs and the outputs and then work to optimize that's what I've heard reporting on Amazon a bunch of other companies and it seems like that's what's going on in the marketing div division on your generative AI products correct very interesting okay let's talk about uh operational efficiency so you've used this to to automate support in some ways can you tell us a little bit about that yeah so we've recently rolled out the our chatbot that we call Penny to deal with level one level two questions that our customers have in the post booking process and you know this was usually handled by humans or by you know a third-party technology that quite frankly didn't learn as quickly as the large language models do and we've seen some really really good operational efficiency in in in customer questions being handled in an automated way using the chatbot um I can't share the the exact results but we're seeing significant operational efficiency and savings by using this technology as part of our post-booking flow and have plans to expand it to all of our product lines in the foreseeable future super cool all right Marty brre is here with us he's the chief technology officer of Price Line talking all about how generative AI is going to change TR uh change travel and you know another way that generative AI is really going to change almost every organization that has a tech function is code right code completion testing um you're the chief technology officer at the company I'll tell you Marty one of the things that I found is I speak to a lot of people who are running teams of Engineers and they talk a lot about how um GitHub co-pilot and things of that nature are magical and then I say can have you been able to measure a productivity a concrete productivity Improvement in your team and they say well we're we're still getting to that so you know I have a chief technology officer of of a major company sitting before me who's working to put this into practice and I'm going to put you uh on the spot with that question and say how's that going for you um and have you been able to see those results in a concrete way so we've just recently completed a pilot of a product uh that played in the generative AI space as it pertains to code Automation and quite frankly we we had mixed results we had certain pieces of functionality that were very productive productive for our uh our Engineers such as quality you know automation of you know you know unit tests or suggestions in code so those were very positive on the mixed bag results side uh one of the Striking things that we saw was was a developer ready to use the code gener ated by the AI model as a commit into production and there was still some hesitancy around doing that um and I think it's really two factors number one really trusting the technology to complete a piece of code that you're responsible for and commit it to production and two it's getting developers used to working in a pair programming model where the other person you're working with is is a large language model which some developers have not wrapped their head around yet so um you know we've seen mixed results um and I think the jury still out on the proliferation of using this capability to fully automate what a developer does so then talk a little bit about you know the positive side of those results and where you can really see the productivity gains in the future if if kind know if engine down on it I think really the development of unit and functional testing is a GameChanger for say more about that because L large engineering organizations commit a lot of resource to you know quality engineering which is writing these tests in in the form of headcount uh and if you're able to free up those quality Engineers to actually be physically writing code versus writing unit and functional tests you're going to see a significant performance U and productivity gain in your engineering organization by you know deploying those those quality you know or estat Engineers to more of a pure programming model you know two is time to Market if you're able to automate you know these tests the time to market for shipping code will only increase and that's where I see the real value in generative AI is unlocking a lot of the maintenance and support functions that developers have to do and having them solely focus on building great features for your product right it seems like it just makes an organization more inventive if you can minimize some of those tasks that don't necessarily lead to the creation of the product but still have to be done correct so Marty you were just at uh Google Cloud next I think Priceline had something like eight sessions at at that uh conference um there's we we talk a lot here about how the different tech companies are battling against each other right you have like the big conversations about Google and open AI Amazon um Microsoft and I think one of the things that's interesting people don't fully appreciate is that this doesn't have to necessarily be Zero Sum that you could take a little bit of each technology provider and end up building pretty robust uh offering you know for whether you're doing something like code automation internally or providing a chatbot to Consumers or whatever it might be creating marketing content so I am kind of curious if you could tell us a little bit about your experience at Google Cloud next and how you evaluate the um you know the current state of of what these platforms are doing well I think in general you know the the startup Community uh AWS open AI Microsoft and Google they're all doing phenomenal jobs in terms of producing meaningful products that companies like Price Line can can benefit from and as part of our architecture we're using AWS open AI Google and startup companies as part of our technical architecture and they all have a huge part to play and are playing a huge part in our our e-commerce platform as far as uh what I saw at Google next I was I was stunned by the level of innovations that are coming out uh in the gener of AI space that are me can have meaningful impacts to companies like Price Line you know a great case and point is I I saw a company on the floor that automates the creation of customer segments and audiences for for marketing and audience creation for a company like price line is super helpful when you're targeting specific deals and values for certain customers typically that would take a a cycle of engineering time now you can do it in a very declarative way which is which is super helpful uh we saw you know I saw another startup company at googlle NEX that was you know providing you know automated ways in which you can manage pus across your vertex pipeline you know that's super helpful in terms of of cost management so I'm I'm really you know super excited about the level andur of innovation that's happening in these large companies but also in the startup ecosystem which is just fantastic right and talk about talk about audience segmentation you know it's so interesting as I started to learn a little bit more about the type of customers that come to Price Line you know it isn't always necessarily like level of wealth or geography or gender or age but it is sort of like what people see as a good deal um you know whether that's they want the you know a deal in a five-star hotel they want a cheaper flight the more direct flight I mean can you talk a little bit about how um when the way that you think about audiences and how they how they're you know not necessarily the traditional type of audiences that uh folks might think about in traditional marketing campaign and then how you can use technology to really speak to each one yeah I think you know my view on this is that we want to drive onetoone personalization uh personaliz has been this term that you've bucketed a lot of different people into like like large segments of people get thrown into this tier and this tier what we're really striving for is really onetoone personalized a Ono one personalized product experience because the same person that you could put into an audience the next time they come back to Price Line could be looking for something completely different you know a great case and point is you know I'm U I have a family of four so sometimes when I use the platform I'm just traveling as an individual like for example I'm going to go visit my son at College you know in a couple weeks and it's just me traveling and so the the points of interests the kind of Hotel I want to stay and the kind of flight I would want to stay or want to take is completely different than another trip that I'm planning on uh during Christmas which is going to Barcelona with my family the complete context around that trip changes I want to stay longer the quality of hotel that I want to stay in is probably higher um I may do some day trips out of that so planning some individual flights to and from Barcelona is completely different so my experience on the platform based on what I'm doing and it at the time and what I'm searching on should dynamically change on on who I am so you know T you know I I used to work at fizer and we had always had this notion of creating personalized medicine for the individual and you know I apply the same kind of principles as the price line is we want to provide a very personalized onetoone experience with the consumer based on what their needs are at that time and what they're planning on doing on the platform and that given session yeah it makes life interesting and I guess kind of fun for chief technology officer Price Line fun very fun well thank you so much for sharing all this great information with us today I think I have a much better I know I have a much better understanding of how deta of AI could be used practically in the travel setting and how it's not abstract anymore it's starting to get rolled into production so really appreciate that and thank you so much for joining Marty thank you for having me I appreciate it