S2, E17: Exploring How and AI Can Enhance (or Complicate) Your Travel Plans

Have you ever relied on for your travel arrangements? In this week's episode of Unpacked, we delve into the technology, its current benefits, potential pitfalls, and what it could mean for our future adventures.

Transcript

Aislyn Greene, host: Fasten your seatbelts, fellow adventurers! Join us on a unique journey as we unlock the power of and awaken the explorer in each of us. In this intriguing episode, we’ll traverse the complex landscape of travel planning—

[Record scratch]

Let’s hit the reset button. I’ll give you two chances to guess who—or rather, what—crafted that truly awful introduction. That’s right, , the star of this week’s episode.

I’m Aislyn Greene, and welcome to Unpacked, the podcast that delves into one challenging travel topic each week. Today, we’re exploring the exciting possibilities of in the realm of travel. Despite the cringeworthy writing, this technology has remarkable potential.

Like many others, I’ve experimented with . However, we wanted to take a closer look at its implications for the future of travel. I believe we have the ideal guide for this exploration.

Chris Dong is a travel planning expert and a writer who covers a wide range of topics, from airlines to hotels to loyalty programs. He’s been cautious about relying on a bot for travel arrangements, and he’ll share his reasons shortly.

Thus, he set out on a quest for knowledge, engaging with AI specialists and travel researchers to uncover how can assist us in planning our trips—and where it falls short—and what lies ahead for us.

This episode is both enjoyable and thought-provoking, and I’m confident it will awaken the adventurer in all of us.

Chris Dong, host: is everywhere! Unless you've just crawled out from under a rock, you’ve probably heard about it. In fact, it's the fastest-growing consumer application in history. But are you really ready to entrust your travel planning to it? Honestly, I have my doubts. As a travel writer, I've mostly looked the other way since its launch late last year. I won’t lie; I felt some anxiety about what (or who) it might displace.

However, I was recently assigned to write an explainer about ’s effects on travel for Dinogo. I decided to take this chance to engage with our future robotic overlords. Just kidding! Think of this episode as an introductory guide to for travel, with me as your somewhat reluctant guide. While this podcast is focused on travel, 's influence reaches across various fields. We’ll explore what this means and consider some broader ethical questions.

So, what is ? It’s a large language model, or “LLM,” a type of artificial intelligence specifically designed to understand and generate human language. The term “generative AI” arises from this, as and other LLMs can learn from examples and create something entirely new based on that knowledge. It’s pretty fascinating stuff.

Hannah Mieczkowski: Large language models like are essentially AI systems that learn—if we can call it that—from vast amounts of text, mostly gathered from the internet. Specifically, operates as a chatbot built on a large language model, which means its purpose is to respond in a conversational manner, similar to how you would interact with another person.

Chris: That’s Hannah Mieczkowski, an AI expert with a Ph.D. in human and AI interaction from Stanford University. Hannah pointed out that AI has existed for years; it powers chatbots, shapes your Netflix recommendations, and even enables facial recognition on your iPhone. Yet, for many of us, engaging with AI seemed reserved for those with computer science degrees or programming skills. Then came along and...

Hannah: ...really blew that out of the water for sure.

Chris: Hannah’s research focuses on practical ways we can use AI in our daily lives. Even she was taken aback by 's rapid adoption. By January, it had reached 100 million monthly active users—just three months after launch. For comparison, it took Instagram two and a half years to hit that milestone, while TikTok achieved it in nine months. The appeal lies in its simplicity; anyone can create a free account, and the interface is incredibly user-friendly. For those of us millennials, think of it as a throwback to A-I-M (yes, AOL Instant Messenger) with a robotic twist. You type your question, wait for to process it, and receive real-time responses. Even better, interacts conversationally, making it feel like you’re chatting with another person.

Or sometimes, a travel agent. But before we dive deeper into , let’s pause for a moment. For many of us, the travel planning process typically begins with one essential tool: Google.

Seth Borko: Most travel planning and processes kick off with online searches or in some way involve searching online.

Chris: That’s Seth Borko, a senior research analyst at Skift and the author of a report from April 2023 titled “Generative AI’s Impact on Travel.” Just so you know, he was stuck at an airport during our conversation, so you might hear some interesting background noises. In his report, he identified four ways that generative AI will transform travel.

The first is operational efficiency, which allows travel companies—like airlines or hotel brands—to maximize their output and work more intelligently. The second involves customer support, envisioning chatbots that function more like real customer service representatives. The third is reputational management, which entails analyzing reviews to help businesses like hotels and restaurants improve, ultimately enhancing your experience. The fourth, well, relates back to how we conduct our searches.

Seth: The fourth aspect, which will take the longest to develop but is also likely the most significant, pertains to search. Right now, when people want to book a flight, they go to Google and ask, “What are the best beaches in the world?” and Google provides a list of websites. Or they might ask, “What flights are available from New York City to Miami?” and get back a list of options. In the future, you could go to and say, “Hey, , what are the best beaches in the world? Can you help me find a flight for under $500?”

Chris: That’s the future, and we’ll explore it further later on. But what can it do right now? Hannah, our AI expert, says that currently, it excels in the initial stages—you know, when you're just starting to dream about your next trip.

Hannah: I find that these large language models are quite effective for ideation tasks. For instance, you might think, “I want to go to France this summer; let me explore different regions and activities there.” You could prompt with something like, “Let me know what’s out there!”

Chris: In other words, isn’t quite the ultimate travel planner in its current state. Rather, it serves as a valuable tool to enhance our research. You’ll still need to make the final decisions.

Hannah: I believe it’s great for giving a broad overview of what people might be interested in, but humans excel at narrowing down choices and focusing on the finer details.

Chris: About a month ago, I created my account for the first time and began asking it questions. I discovered that the more specific your prompts are, the better the responses. For example, when I inquired about peaceful, phone-free retreat options, it suggested Bali and Tulum, among other well-known destinations. However, when I requested a less conventional spot, it recommended Dharamsala in India, a town nestled in the Himalayas and home to the Dalai Lama. Pretty cool!

Currently, shines in responding to more straightforward queries, such as “What are some great travel destinations that don’t require a passport?” or “How do I get from Tokyo Narita Airport to Shinjuku?” In contrast, when asking for a three-day itinerary in London that’s suitable for kids, might overestimate what a human (and a little one) can realistically accomplish in a single day.

Questions that are highly specific, like “What are the best family-friendly hotels in Hong Kong for under $300 a night, within walking distance to local attractions?” are on the cusp of being reliable, but they still face challenges due to a limited number of data points and information only extending through 2021 in the free version.

In summary, the quality of answers can vary from somewhat decent to quite poor.

However, the responses are likely to improve as more travel companies get involved and, crucially, provide data to .

Seth: We’re witnessing major online travel agencies experimenting with this. For instance, Expedia.com and Kayak, a subsidiary of Booking Holdings, have already introduced plugins for that essentially supply it with real-time travel data from their extensive databases of hotels, flights, and activities. This way, users can engage with generative AI while receiving up-to-date travel-specific information.

Chris: In April, travel giant Expedia started testing a new in-app travel planning feature powered by . You can ask questions, like where to go for a family Caribbean vacation, and instead of being redirected, you can actually book your trip directly within the Expedia app.

Then there’s GuideGeek, a new -powered travel assistant from the Matador Network that provides travel tips along with live flight and vacation rental information. Even cooler, it’s all available through WhatsApp. Many other companies are developing similar tools too.

Naturally, travel brands are excited about all of this, as it enhances (and sometimes replaces) human interaction, ultimately reducing costs. As noted, Expedia aims to connect the idea phase with actual booking. Rahti Murthy, the company’s CTO, mentioned that consumers can chat with to brainstorm their next trip, and when they're ready to start looking for flights, hotels, and more, they’re already in the app. Think of it as a comprehensive one-stop shop.

Well... almost. There are some significant drawbacks to all of this at the moment, too.

OpenAI claims that -4, the latest iteration of the chatbot, is capable of creative thinking to tackle complex problems. However, it has substantial limitations, such as potentially outdated data. Additionally, there are numerous ways it can mislead users entirely. LLMs often exhibit what researchers term “hallucination,” which means they can generate information that appears factual but is completely incorrect. Like, really wrong.

Hannah: They aren’t the most reliable source of truth. Therefore, I recommend seeking out additional sources of information, even if just to verify what the model has produced. It’s generally a good practice to confirm information you receive from both people and AI systems.

Chris: Hannah recounted an example from her own experience using to understand how to utilize travel points from her Chase credit card.

Hannah: Earlier today, I was experimenting with it, asking a few travel-related questions. I mentioned, “I have 80,000 Chase points and I want to visit Europe, what should I do with them?” The first suggestion it provided was to use the Chase Ultimate Rewards portal. However, since I follow various points experts in online forums, I knew that might not be the best route. I could verify that information myself, but if someone took 's response at face value—assuming it was the optimal choice—they might waste valuable points unnecessarily.

Chris: There’s also a disclaimer prominently displayed on ’s homepage stating, “While we have safeguards in place, the system may occasionally generate incorrect or misleading information and produce offensive or biased content. It is not intended to give advice.”

Ultimately, this is why Seth believes AI will never be a standalone solution.

Seth: Even now, online searches are often paired with other resources, such as social media influencers, recommendations from friends and family, and travel agents. I don’t think it will function as a solo recommendation tool. Instead, it will coexist with other options, and achieving accuracy is crucial, as travel is a significant investment for most people, often a once-a-year or once-in-a-lifetime purchase. The stakes are incredibly high.

Chris: Despite everything, the ongoing advancement of and AI could dramatically transform the travel industry.

For travelers, the prospect of enhanced planning and booking features is exciting. However, there are still critical questions surrounding the technology and its potential impacts on travel that remain unresolved. The inputs for are not explicitly defined, which opens the door for manipulation of the information it provides. Without proper safeguards, responses could easily be skewed by those who seek to exploit the system.

Seth: There’s a whole industry focused on search engine optimization aimed at ensuring that your content appears prominently on Google. One can only speculate that a similar situation will arise in the realm of generative AI, where individuals will strive to ensure their hotels and destinations are highlighted in and other similar platforms. It raises a crucial question about what criteria will influence ranking in 's search results. Rest assured, there are already numerous individuals trying to decipher that puzzle, and they’ll likely market their expertise to hotels, tourist spots, and businesses both in and outside the travel sector.

Chris: He likened it to the chaotic early days of search engine marketing.

Seth: Remember when there were those circular networks, where everyone would link back to one another because they understood that backlinks mattered in Google’s algorithm? Then the system evolved, and SEO became more sophisticated. We’re likely heading into a similar scenario again.

Chris: Currently, there’s no regulatory body overseeing this type of technology, but Seth believes that rules will eventually be established.

Seth: This question transcends just travel; it delves into regulatory and political realms. It prompts us to consider how we view algorithms, their operations, and which ones we can place our trust in.

Chris: This raises concerns about manipulation and deeper issues. , as a nonhuman entity, doesn’t factor in sustainability or broader ethical travel matters. What implications does this have for emerging destinations seeking recognition, or for those wishing to avoid mass tourism spots? Ultimately, those who excel at managing inputs will prevail, much like navigating a Google Search today.

Seth: Essentially, AI tools can only suggest what's already available online, which tends to favor popular tourist spots. So, if you ask about the best beaches, expect answers like Maui, Cancun, or Phuket. This isn’t merely about accuracy; it raises concerns over bias towards well-trodden destinations, overcrowding, and the fate of lesser-known places trying to gain traction. What if you desire a more unique experience instead?

Chris: While we lack definitive answers right now, one concern we don’t need to entertain, according to Hannah, is sentient AI.

Hannah: There's significant hype surrounding and similar large language models across various fields. People might be exaggerating their capabilities. While future advancements are possible, I advise caution against fears of singularity or AI domination.

Chris: Ultimately, the need for humans will remain essential.

Hannah: I truly believe AI should serve as an enhancement to human abilities. To elaborate, I don’t foresee a time when these AI models can operate completely on their own. Human involvement is crucial at every stage, from training to implementation. Unless there’s a significant shift in the paradigm—which I don’t expect anytime soon—this reliance on people will persist.

Chris: My exploration of and its impact on an industry I cherish is just starting. As an enthusiastic travel planner who values personal recommendations, I can’t imagine using to plan trips or discover activities in new locations. However, I can see its potential in addressing specific issues, like the common frustration of dealing with customer service.

Hannah: While progress in this area has been slow, there are AI models emerging that can assist with tasks like checking in for flights, making reservations, or resolving customer service problems. These systems can access your information and manage these tasks for you, easing your burden.

Chris: The larger question looms: Will it lead to a near-utopian travel experience or a perilous reality where it’s hard to tell fact from fiction? Honestly, I think the truth lies somewhere in between—at least, that’s what I hope.

Aislyn: I totally agree, Chris. For those interested in diving deeper into travel and , we’ll provide a link to Chris’s article for Dinogo in our show notes, along with the Skift report Seth referenced earlier. You can discover more of Chris’s work on his site, thechrisflyer.com, and connect with him on social media at @thechrisflyer. For loyalty points enthusiasts, Chris also offers consultations for travelers looking to maximize their points for trip planning—details are available on his website.

A big thank you to today’s guests, Hannah and Seth. To dive into Hannah’s work further, visit her website at hnmiecz.com or follow her on social media @hnmiecz. You can find Seth at skift.com or connect with him on social media at @SethBorko.