PROS is excited to launch our new podcast, Ahead of the Curve. Check out our first episode where we chat with PROS' own Michael Wu about the exciting future of AI in the travel industry, and stay tuned for future episodes!
Surain Adyanthaya: Welcome to the latest episode of Ahead of the Curve. I'm very excited about today's episode, I think we're gonna touch some very interesting and important topics for the airline industry today. Our guest today is the Chief AI Strategist of PROS, Dr. Michael Wu. Hey Michael, welcome....
Michael Wu: My pleasure to be here. Yeah.
Surain Adyanthaya: Good, good. Well, to start it all off, can you introduce yourself to the audience, you've had a very important career and contributed in many ways. Let us, just talk about what you've done a little bit.
Michael Wu: Well, okay. [chuckle] I'm not big with introduction, but sure, I'll give a quick intro. So I joined PROS about like three years ago, basically just focusing on leveraging AI to help our business be more competitive and be more efficient, and also like work with scientists to essentially improve our algorithm, leveraging more data sources and some of the latest and greatest from the Silicon Valley. And a little bit of a background, I've been in the Silicon Valley in startup scene for about 10 years doing social media analytics, big data, machine learning and all that AI stuff before AI and machine learning was popular. It wasn't even called AI and machine learning back then, it was just called analytics and applied statistics. So, [chuckle] you can see how long ago that was, [chuckle] so...
Surain Adyanthaya: Yeah, for sure. Yeah, and that's... So much has happened in AI and it's become quite the topic that we hear about all the time. So I guess to just level set for all of us, given your experience, can you give us a simple definition of what is AI and how does it work, how should we think about it?
Michael Wu: So, there's actually two questions there. [chuckle] So there's a... What is AI and how does it work. Okay, so the first part, so the simplest way to look at AI is really think of it as a machine mimicry of human behavior with two important characteristics. One is that it has to be able to automate human decision and action. And the second one is that it has to able to learn, that means you have to be able to improve its performance over time with usage. So that's it, that's what AI is, if you actually meet these two criteria, you can automate human decision and actions, such as driving a car or diagnosing medical disease or whatever, so... And you can actually learn and improve, and do better at those tasks, then it's a AI.
Michael Wu: So, now that we know what AI is, we can talk about how it works. So, it really also divides like a two step, the first step is really learning from data. Learning from data, which encapsulates past human decisions and actions, and so this step is typically referred to as machine learning but essentially we are just trying to turn the data that... The data asset that most company have now, into some kind of a model or some kind of algorithm, which they could reuse over and over again to mimic this human decision and action from the past. So once you have that you can automate human decision, you need to have the second component, which is learning, improving over time. And that part involves, essentially getting the feedback data to improve the model. And this is, sometime I call this the learning loop, essentially it's what makes AI smart because every time you make a decision or took an action, you get some feedback and the AI actually learn, did it do well or did it not do well and they actually refine this model to be a better model, more optimal model, so next time it actually makes a better decision or takes a better action.
Surain Adyanthaya: Yeah. That totally makes sense and I guess in line with what you're saying, data has become much cheaper to collect and store and with cloud-based computing, computational power has become much more accessible and commoditized for all of us, all the companies. Has that helped AI in its evolution?
Michael Wu: Yeah, definitely, I think that's actually what spawned so many, I would say, startup, which are, I would say, resource constraint [chuckle] to jump on this area, and I think the accessibility of this technology and availability of this big data and also kind of open source algorithms have made this area just explode.
Surain Adyanthaya: I guess we've seen AI, as you mentioned, in many different ways with autonomous vehicles, etcetera, it gets a lot of attention in the media, but there's a great potential for AI that's already at work in travel and has many potential use cases in the future. Can you talk a little bit about how AI can change the future of travel?
Michael Wu: There's a lot, [chuckle] literally, I could go on for days on this topic, but you could kind of take a look at it from two perspectives. One is from the consumer's perspective and that's basically, AI is gonna essentially change on the future of travel experience for travelers. And so that's gonna be a very big dramatic shift in the coming days. But also AI can change how, I would say, airline or carrier operates, to be more efficient or... And so that would change, greatly change and improve the business operation of the airlines, so... Yeah, that's just a lot, so... [chuckle]
Surain Adyanthaya: Right, okay, great. Well, I think maybe in a future episode, we can talk more about the business side of things. For now, let's focus on the consumer, on the traveler himself. Let's talk a little bit about how can AI sort of change the travel experience for the consumer for the better?
Michael Wu: Let me answer that question with a question back to you. [chuckle] So, you are obviously a global traveler, a frequent flyer yourself, what is your biggest frustration about air travel?
Surain Adyanthaya: I guess for me, the frustrations can be, I hate wasting time, waiting in line, I like transparency about my travel that's upcoming for sure, but any delays, disruption in my travel plans really are taxing and I would love to reduce or avoid them.
Michael Wu: Yeah, exactly. Same as me, I hate waiting in line as well, and I hate... Obviously, delays are always a headache for... Especially for frequent traveler, if you travel so much, little delay adds up, and here's the thing, so airlines are trying to fix these delays or disruptions, it usually end up putting people in-line, the real benefit of having AI to help this is that it can actually alleviate this type of experience, it can automate a lot of decisions or actions that normally is done by human, so AI is actually best for automating repetitive tasks done by humans.
Michael Wu: So think about all these delays or disruptions or waiting in line, what is the person that are doing? They're doing pretty much the same tasks over and over again for different people. So if you have a AI that can actually help them automate some of those tasks and some of those decisions, then they can actually go a lot smoother, a lot faster, 'cause you don't need a human there, the agent there to kind of help you anymore, you can actually do this through a self-serve kiosk or something like that, which you can install hundreds if you want to. So for example, face recognition for check-ins or for bag checks, or maybe even border control, so these are things that could... You can have hundreds of these kiosks, and rather than having... Humans are kind of right now the bottleneck at this point, so I would say that this is one area that certainly would change dramatically in the future with advancement of AI.
Surain Adyanthaya: I know that just about every customer that we work with, the customer experience is extremely important, and airlines are always looking for ways to improve the customer experience, make it less... Reduce the friction of the experience, make the journey much more pleasurable, so it sounds to me like you're saying AI can help in many of these touch points in the traveler journey. It sounds...
Michael Wu: Yeah, totally yeah, I would say that some other technology such as chatbot, it's used in the consumer world a lot already today. There... I would say that chatbots are great as a AI, as kind of a perceptual AI, but it's... I would say, you can use it to make people's life more efficient, but you could also add another layer of kind of personalization over it as well, so it becomes... For example, it doesn't always have to respond to you in English, it could respond to you in your native tongue, so we have a multi-language chatbot that understands where you're from, where you're going, and then can basically be a virtual agent that you could call on any time, you don't have to just look for agent, it'd be on your phone.
Michael Wu: And if there's any, I would say, delays or cancellations or anything, it would be proactive to kind of respond to your need, or most people when there is a delay or cancel, they need to probably reschedule, the system already know where you're trying to go, where you're flying from, so why not just give you... Send you all the choices and based on your past travel behavior, give you a rank list of maybe you like to fly in the morning or sort them just in some way and let you choose, and if there is a cancellation you need to stay overnight, why not just distribute these hotel vouchers automatically to your phone, so you could... If you use it, great, if you don't use it, [chuckle] good for you. So all of this can be automated without human touch at all, so... [chuckle]
Surain Adyanthaya: For sure. And I guess you touched upon an important point that I hear from our airline customers all the time, it's about personalization, knowing the customer and offering something that's extremely compelling, transaction-able, providing value for that consumer. Can you talk a little bit about personalization in AI?
Michael Wu: Yeah, yeah, definitely. So personalization is definitely one big area, probably one of the earlier area of AI development, as I said before, all these chatbots and maybe even augmented reality, they can give you guidance, how to navigate the airport and everything like that, they can be used just as a technology by itself, okay, without AI or... It's not being smart at all, you could use it just as a digital signage or to show you... Maybe translate the sign from the language that you don't understand to the language that you understand. So all that can be done, but if you add another layer of personalization to it, then it just... It becomes much more, I would say, helpful.
Michael Wu: So it goes essentially from being a tool that provides efficiency to being hyper-personalized, it's actually trying to solve problem that you're trying to solve, and I would say that a lot of this personalization, probably it's a lot more important before even the passenger gets to the airport, [chuckle] it's actually before 'cause probably before they actually wait in line, and when you actually wait in line... Yeah, there's certain levels of personalization they can do to help you specifically, so that the... Right now... Previously when I talk about automated distribution of hotel voucher or scheduling, so these are probably... Apply to a large group of users.
Michael Wu: Little difference, they know this plane is cancelled or something like that, this plane flies from A to B, so it just gives you all the lists, this different itinerary that you could choose from, so you'd probably do that for all the people who is impacted by the same, but how does that actually add another layer of personalization. Is because they can actually look into your past travel behavior, maybe, like I said, if you fly in the morning or if you like to get home before five or something like that. So these are behavior data that the system can actually leverage to help you provide... Maybe recommend, you probably like this flight the best, then... Yeah, it's actually much more conducive for people to click on that then move on, so...
Surain Adyanthaya: For sure. And as we all know, the traveler experience begins way before the actual day of travel, and that pre-departure experience is extremely important, it's perhaps the most stressful part of the journey for many people. So, can AI help improve this sort of pre-trip traveler experience?
Michael Wu: Yeah, obviously. I think that... That's actually where, again, I wanna reiterate that personalization is actually more important, even all the way at the beginning, I would say the marketing phase, people can actually generate, I would say, personalized content, images and text in real time, so they actually generate image that's actually interesting to you. Maybe they know that you like a certain type of destination or a certain country, then they will show you those images, and to make it more conducive for you to look at it. And you watch with GPT-3, how to generate text messages to go along with those images, so these become hyper-relevant, so it's no longer... Marketing is no longer, I would say, creating content that are kind of prebake in some way, they are kind of designed with a bigger audience to... But keep in mind, when you have this for a bigger audience, it's no longer specific to you, so when you actually could create a piece of micro-content, a tiny piece, a little message, a little image that goes with it that's specifically for you at that moment, then it's actually very compelling.
Michael Wu: The action it's able to drive is actually much, much more compelling. So, from marketing, to shopping, I would say, personalization, obviously could recommend bundles or... Since they know your travel behavior past or what kind of ancillary you might use, or take advantage of, and maybe even offer, I would say, a complementary type of promotion for car rental or hotel depending on what you need, maybe you're the one who never go through the car rental, then they won't recommend you that, maybe you're the one who like to sleep and don't like to eat on the plane, then they won't recommend you that. So, they know that you're very interested in doing work on the plane, then they will give you the Wi-Fi offer, so these are things that are definitely relevant and specific to you, or maybe they will figure out that only when you fly in a long-haul flight you like to do that, but for a short flight you just wanna take a nap, then for the long-haul flight they'll recommend the Wi-Fi and for a short-haul flight, you just... Recommend you, not very much, so give you a good night's... Good rest, so... Yeah, there's a lot. Yeah.
Surain Adyanthaya: For sure, and I think we're already starting to see some of this AI technology bearing fruit. I know at PROS we've been working on some of this for a while, and it's very exciting to see this next phase of AI really helping improve the customer experience, it's very exciting to all of us I think. And I guess the final question or topic I have for you, Michael, is... I'm of the camp that COVID has changed the world forever in some ways, how can AI help improve the travel experience in sort of this post-pandemic world that we're entering today?
Michael Wu: In the post-pandemic world, obviously, we are going to a more kind of touchless, paperless kind of world. You don't want to... 'Cause obviously to prevent the spread of COVID or any variant that may come up, so I think a lot of technology that we talk about before, such as face recognition, chatbots or augmented reality, when used with personalization, people won't need to interact with other human for many, many situations, but if you just have a chatbot or augmented reality that's not personalized... Maybe it's okay, just not what I want. It's good that I can ask a question, you can help, great, but it's not specific to me, so some of them may still go and look for agent or look for someone to help them, but when you actually have something that's very personalized, face recognition, chatbot and augmented reality that's personalized to that person, then you can actually help them resolve most of the issues, so that they don't need to actually contact anyone or get in touch with anyone in-person, so that certainly will help to improving this touchless and paperless world of travel.
Michael Wu: And I would say that, there's a lot of... In the airport or any public place, there's lots of, I would say, video cameras around for security reason, certainly in the COVID time, social distancing is a big thing. You can use that, those video feeds to catch or monitor, are there areas that are kind of too crowded, are people kind of properly social distancing, are they... Or should they send someone there to kind of help direct the traffic so that people are... Stay within the safe distance, or you can even use it to kind of monitor for suspicious objects or suspicious behavior. So... Yeah, so video analytics, it's obviously another area that was certainly helpful in the post-pandemic world, and obviously I think that one area that will be big in the near future in travel is robotics, and I think that robots run...
Michael Wu: Essentially, you can have a robot that runs around that's not smart. [chuckle] And obviously, that's not gonna be very good, probably run into people or... And things like that. So, to add a layer of kind of intelligence to it, these robots need to behave like a self-driving car, it needs to have this autonomous AI built into them to essentially help them have some level of intelligence, to know not to collide with each other, not to collide into people, and maybe stay in the proper lane, so I think robots would definitely help, I would say in the post-pandemic world, there are already, I think, cleaning and disinfecting robots that use maybe UV light or something, and they just go across and then disinfect a whole big region and maybe handling your duty free items and deliver them to the gate, or even handling, tracking luggage, and people hate to be lost at luggage, if you have something that is, I would say, handled by robot, they're probably less error-prone than human in general. So I would say that this is also another area where...
Surain Adyanthaya: For sure. In fact, I think it was in the Taipei Airport, I ran into a robot helping passengers find their connections, which was really interesting, and people were actually looking forward to engaging with it, which was fascinating to me to see that, but I think we've run out of time, thank you so much for joining in this episode, and I'd love to have you come back, this time, this focus on the consumer experience was excellent, but I'd love to have you come back for another episode where we can talk more about the business of airlines and travel and how AI can really be applied in that area, so I hope...
Michael Wu: It would be my pleasure to do so.