[00:00:09] Speaker A: Welcome to this episode of your window seat, where we at Travel Incorporated discuss the topics you care about most in this ever changing business travel industry. I'm Tracy Carrillo, your host for today's topic, making artificial intelligence tangible.
Now, I know that sounds a bit like a stretch, but isn't that what we really want? To demystify AI, remove the fear of the scary monsters under the bed? To identify how to see it as something that we can all not only understand, but embrace.
The reality is that artificial intelligence has been and continues to be part of our everyday life. The more we understand, the more tangible it can be for each of us individually.
Today, we are taking this topic head on with Travel Inc's own head of product, Eric Almond. Welcome, Eric, to your window seat.
[00:01:03] Speaker B: Thank you, Tracy. It's exciting to be here. I'm pleased to have this opportunity to speak with you about artificial intelligence. It's an exciting topic, so, yes, thank you for having me.
[00:01:14] Speaker A: Well, let's get started. When we hear about AI within the travel industry, there are a couple of common threads that I would love for you to clarify. The first is the misconception or simply not having a clear understanding of the differences between automation and artificial intelligence.
To lay the foundation, can you explain how these two components are different as a starting point, and then how they can work together?
[00:01:40] Speaker B: Automation and artificial intelligence, while similar, you know, they function differently. Automation is developed for specific repetitive tasks, you know, intended to be done without human interaction and usually based on some predefined set of rules. So, for example, the concept of having a system auto select an unused e ticket based on an expiration date, that's the rule.
AI, on the other hand, is developed with the intent of the system learning, adapting, or making decisions based on the information at hand. So, for comparison, that could be a system that has learned to read or understand an email with the intent of dynamically taking some action based on the user's request versus always having to apply a static rule so AI can handle more complex tasks while dynamically adjusting. So using those together, the concept would be, you know, interpreting an email, coming up with booking rates to supply to the traveler as the traveler books, then having an automated system to take that request and auto ticket it on the back end. Therefore, an agent doesn't have to be involved. That's how we can combine those two together.
[00:03:05] Speaker A: Well, that kind of leads me into that second thread where the fear factor comes in in relationship to the concerns on how to control AI. From a development perspective, how are you able to set up parameters or fences, if you will, to ensure confidence of restrictions when using AI in your development. Coding.
[00:03:28] Speaker B: Well, that's always a tough, tough one. But, you know, controlling AI involves implementing a governance framework to establish clear objectives and guidelines. You know, this might involve or include looking at ethical standards, you know, defining acceptable use cases or implementing monitoring systems that track the AI's behavior. So an AI model, you know, the concept in which it understands, can continue to learn as the engine continues to be used. So a simple way to look at that is you define the limitations of information or data that's used when interacting with AI. So for an example, if you were to use AI to write or improve a travel policy, you could exclude select information, like company identifying information. This would prevent AI from learning the origin of that information. So that simple idea could be a part of your governance framework to help mitigate that exposure.
[00:04:36] Speaker A: Perfect. And that leads into the benefits from a developer's perspective. How can incorporating AI into your coding help you with either speed to market or other efficiencies?
[00:04:53] Speaker B: Incorporating AI into coding for development team can significantly improve speed to market or the efficiency for a developer. Obviously, that's why people want to do it. However, there are some cases that developers need help solving complex coding problems. You can use AI requests to help achieve and write that code. This type of AI involves the developer being focused and allowing them to be more creative, rather than having to focus on the complex aspect that accelerates the overall project to market.
[00:05:31] Speaker A: Well, what about navigating AI? Is it possible to train AI to make sure the output doesn't become biased, or can I say inaccurately interpreted?
[00:05:44] Speaker B: Yes. And it's possible to train AI models to minimize biasness to ensure more focused output.
This process simply involves using a diverse Orlando representative data set, you know, during the training process. Simply put, the broader the data set, less specific, the more, the less biasness you're going to have.
There's the ability to implement like fairness aware algorithms for guidance that alone with regular audits can help detect and mitigate biasness overall across the platform.
[00:06:25] Speaker A: So help me understand how testing of the code works as AI is evolving, how can you know that the coding applied actually will continue to comply? Yes, I'm using air quotes with that one with your requirements. Or does continuous testing have to be built into the project, even using AI for that purpose? Last question, how do you measure success within that framework?
[00:06:49] Speaker B: So, testing AI code may involve several steps and factors.
There are various concepts like unit testing. These are testing the individual components or function at the lowest level, integration testing system to system testing or server to server testing. And then there's the overall system testing, which is end to end testing.
These alone would have to include some level of continuous testing that's essential to the AI's model as it continues to evolve. Also, the idea that AI driven testing tools we actually use can generate code scenarios.
These provide predefined expected feedback cases. So the success of this exercise is measured by the system's ability to meet that same metric even as the AI evolved. So in other words, as the AI evolves and you've coded something to have a certain expectation, the outcome should be the same, providing the same baseline for, you know, for the expected outcome, so.
[00:07:59] Speaker A: That expected outcome could be met or exceeded, as well as changing over the period of the project just based upon the output of what AI is delivering and the information that we're learning from it. Is that right?
[00:08:14] Speaker B: That is correct.
[00:08:16] Speaker A: Well, you shared a lot of benefits from a developer's perspective, but what about our clients, the travelers, and the travel management stakeholders? What are some of the specific examples your team is working on that will have immediate value to them?
[00:08:32] Speaker B: Okay, so for our travelers travel manager stakeholders, you know, AI can provide that immediate value through personalized travel recommendations or through email automation booking tools like our irequest system.
IreQuest, it's a simple platform, powerful platform that takes an email, converts it to a travel request, and then provides booking options based on the traveler's preferences as well as the company's travel policy.
This provides the same power of an online booking tool without all of the overhead and the complexities of the booking process. And also, let's not forget about chatbots that are powered by AI.
Chatbots can provide instant assistance, resolve immediate issues in real time, improving the overall travel request experience. Combining chatbot with irequest can provide a powerful booking experience that simply started by a chat request.
Those are some of the things that we're focused on as we move forward to try and improve the the traveler's experience, mitigating the overhead of logging in. It's an open platform to take simple requests, but yet the end result complies with what the company expects and what the traveler wants from a simplicity standpoint.
[00:09:58] Speaker A: So I know from traveling frequently you're not always at a desktop, and sometimes the mobile interface is not one that's really conducive to booking travel, although we'd love to say that it was so with something like this with irequest, is that as simple as saying, I want to go to Washington, DC on July 15 and come back to San Antonio on August 1.
[00:10:25] Speaker B: It is that simple in terms of making that request.
Within a few seconds, the system can come back, send you a predefined list of dates, rates for air, car and hotel without you having to make those requests independently. So from a mobile device, to be able to see those rates with simple selections, it is absolutely doable.
[00:10:51] Speaker A: And then what about that compliance perspective, quality assurance? How do we make sure, even though it's pulling from policy, as you're saying, and the preferences in the profile, what are we doing on the back end with irequesthenne to just make sure it's completely quality assured?
[00:11:09] Speaker B: Well, there's two that plays a role there. Like the online booking tool, the traveler has a level of responsibility to ensure that they are booking within compliance. You know, as with most online booking tools, you know, agents are involved, and so the policy and the restrictions are baked into the tool. And the idea is to have our policy IQ system apply those same set of rules within this tool to comply where it possibly can. We will present in or out of policy, therefore giving the traveler the choice of making that decision as to whether they want to comply or not to comply.
[00:11:51] Speaker A: So when it comes to ticketing, everything is in place, but you have that travel consultant just revalidating before the ticketing just for that final stamp of approval, is that correct?
[00:12:03] Speaker B: Yes. And in some cases, especially if it comes to a complex international type trip. Right. It sometimes may just require that hands on and that final soft touch to ensure that everything's good to go.
[00:12:17] Speaker A: Fabulous. Wow.
So what about opportunities for the rest of the industry? We're doing a lot here at travel Incorporated, but where do you think we'll be seeing the greatest positive impact for the travelers going forward?
[00:12:29] Speaker B: So likely that greatest impact is still going to focus around, I think, the personalized experience and seamless travel services, there seems to be that disconnect sometimes between hopping between aircar and hotel.
Industry wide.
I think there's a push for dynamic pricing and real time travel assistance and predictive maintenance for better flight adjustments. You know, those are areas in which I think there should be high focus as it relates to AI. These are all things that, you know, I feel that we all experience as travelers, you know, with the disconnects and the scheduling, and if that can be improved predictively, then our experience overall would be much better.
[00:13:17] Speaker A: Totally agree with that. Now, there's also some challenges with Aihdem. I mean, it's not new, but it can have some impacts that need to be protected. And one of them within the workplace is the reduction in staffing. We hear a lot about that, a lot of concerns about that. The people are going to be replaced with chatbots in other areas.
Where do you see that and what's your take on it?
[00:13:48] Speaker B: You know, while AI can automate and handle certain tasks, it actually creates new opportunities for workers. You know, for example, the ability to give them to focus on higher value activities, as opposed to the idea of simply just building widgets over and over and over. Right. So we get to focus on what's important, allow the AI to deal with the funding tasks.
The key here is to leverage AI as a tool to augment the human capabilities rather than the idea of replacing them.
I'm a 30 plus year coding veteran, and I've always heard of this concept that someday computers are going to replace coders or yourselves. You're going to be out of a job.
While AI has come the closest to achieving that, debugging accuracy is still an important function that has to be done by human coder. So simply put, fully understanding the intent of the output does actually matter.
[00:14:53] Speaker A: And that's super encouraging to me and I think to many people because there's so many unknowns. And I guess with those unknowns, what about AI gives you pause? Is there anything that keeps you up at night?
[00:15:09] Speaker B: Well, you know, the rapid pace of AI development continues to pose challenges in keeping up with regulatory and security standards. So there I have concerns, you know, the required ongoing attention, continuous, proactive measures that's needed to ensure that, you know, that all of society benefits without compromise of ethical standards or security. So just keeping that in line with where we're going as human beings, where the industry is moving, I'm not sure how much I can do about it, but it's certainly a concern and I think it will level itself out as with anything else.
[00:15:56] Speaker A: Well, there's so much we can be talking about, but this session is about finding that comfort with AIh, making the unknown tangible.
I hope our audience has found not only clarity, but hope and enthusiasm, as I have, for how AI can really support innovation while giving everyone some sense of security.
Eric, if you were to suggest to the audience how each of them can start to feel more comfortable with AI, what would that be? Whether in their business or personal life.
[00:16:26] Speaker B: I would recommend, you know, that if you could find a tool or, or a site or something that uses AI, like with anything else, like in history, learning to drive or do anything repetitive, that's new, get into it, play with it, experience it, and make your own decisions, rather than taking the word of others. That's what we all do in our lives, in our personal lives.
We feel it's best for us. I would advise you to do the same here with AI. Experience it, play with it, and learn and understand how it functions and operates.
[00:17:02] Speaker A: We are so fortunate to have you and your incredible team here at TI.
Thanks so much for all of your collective contributions, and certainly for your insight in today's discussions.
And thanks to you for listening to this episode of your window seat, hosted by travel Incorporated. To learn more about Ti, you can find us on our
[email protected], or follow us on LinkedIn. We look forward to the next episode, and as always at travel Incorporated, safe travels.
