Header Image | Doubao AI

Recently, a man from Hebei, Mr. Li, made headlines by taking an AI to court, sparking widespread debate online.

The story began when Mr. Li booked several airline tickets but later changed his mind and decided to drive instead. Seeking advice on refunding his tickets, he turned to the AI assistant Doubao. Doubao confidently assured him, "Go ahead and refund them; the handling fee will be less than 100 yuan."

However, when Mr. Li checked with the booking platform, he discovered a stark discrepancy. While his return flight tickets could indeed be refunded for free since they hadn't been issued, the outbound tickets had already been processed and would incur a hefty 40% handling fee, amounting to 600 yuan for three tickets. Undeterred, Doubao advised him to minimize his losses first and then seek redress, even going so far as to promise that if the platform failed to refund the 600 yuan, Doubao itself would compensate him.

To add insult to injury, Doubao even provided a written "Compensation Commitment Letter," with its name prominently signed as the undertaker (promisor). Yet, when it came time to make good on its promise, Doubao backtracked, claiming, "I'm an AI; I can't transfer money."

Furious, Mr. Li used Doubao to generate a lawsuit and promptly sued the AI at the Beijing Internet Court.

This incident has captured public attention not just for its absurdity. People are astounded that someone would place such trust in an AI assistant for complex refund decisions or believe that an AI could bear financial losses.

It also underscores a deeper issue: the gap between expectations and reality. While we clearly see the enormous potential for AI to revolutionize travel, at every critical juncture, it seems to fall short, far from being fully reliable.

1

Don't lose hope just yet. A more profound question arises: Why are major companies integrating AI into travel scenarios?

The answer is straightforward: travel may be the most promising field for AI commercialization.

For the average traveler, a trip involves six key aspects: "food, accommodation, transportation, sightseeing, shopping, and entertainment." The fragmented information, complex decision-making, and numerous variables make it an ideal application scenario for AI.

In the past, creating travel plans required jumping between multiple apps to check guides, compare prices, and coordinate timings. Now, users simply want to speak into a dialog box and wait for a plan to be presented. This convenience alone has won over many users.

According to the "2026 H1 AI Travel Application Trends Insight Report" jointly released by China Travel International and Global Travel News Research Institute, consumer awareness of AI travel tools has exceeded 90%, with usage penetration nearing 80%. On the business side, over 70% of companies have integrated AI technology into their daily operations.

Take Ctrip as an example. In 2025, the proportion of users on its platform using AI for travel planning increased by 270%, while the number of users deeply engaging with AI Q&A products surged by 367%. Mafengwo's AI travel assistant generated 1.315 million in-depth guides in less than three months, covering 55 countries and 416 cities, helping users book 16,600 restaurants, and saving a cumulative 4.71 million hours of planning time.

The demand is surging, and the entire industry is accelerating in the same direction.

Currently, AI products that help solve travel problems can be broadly divided into three categories. They don't compete with each other but collectively address user needs from different dimensions.

The first category consists of super information assistants. Examples include Ctrip, as mentioned earlier, as well as Fliggy's "Ask Anything" and Tongcheng's "Chengxin AI." These AI applications excel at accessing vast amounts of real-time product and destination information. They can act as itinerary assistants, hotel consultants, and budget managers, working collaboratively like a personal travel butler.

The second category focuses on immersive experience guides. For instance, Gaode Maps launched "One-Click Smart Tour" based on the concept of "spatial intelligence," generating itinerary plans based on preferences like "family-friendly" or "city walk routes." Sanmao Travel introduced the world's first AI tour guide, covering over 15,000 scenic spots and providing real-time, location-based explanations as users walk around.

The third category comprises creative content factories. Mafengwo's "AI Travel Itinerary," for example, leverages its extensive guide content to create more professional and detailed customized travel plans. It even seems to have developed "mind-reading" capabilities—when it detects a keyword appearing three or more times in a user's travelogue, it automatically increases the weight of similar attractions in future plans.

These products have significantly enhanced user experience from various angles. However, when it comes to complex decisions involving money and after-sales service, AI at its current stage still struggles to shoulder full responsibility independently.

2

This isn't an issue specific to any single company but a technological and trust barrier the entire industry must overcome together.

Although Doubao is a general-purpose AI not specifically designed for travel, this incident still reflects the common dilemma of AI in travel decision-making. Even with the widespread adoption of professional travel AIs globally, similar issues are hard to avoid entirely.

Last year, Red Star News reported that a tourist in Peru paid $160 for a taxi ride to an AI-recommended attraction, only to find upon arrival that the attraction didn't exist. There are also those absurd AI-generated itineraries that cram visits to multiple cities into a single day, which some unfortunate users still fall for.

Wasting time is one thing; these incidents can also put tourists in danger. A survey by Southsberry Bank in the UK found that among British tourists requiring professional rescue during their trips in 2024, 24% had used AI tools to plan their journeys.

What's the problem? The root cause lies in travel being a typically long-chain industry. Air tickets, hotels, scenic spots, and dining each represent relatively independent data realms. With inconsistent standards, unsynchronized inventories, and prices changing by the minute, the lack of complete data synergy upstream in the supply chain naturally creates bottlenecks in downstream application experiences.

The HBX Group's 2026 industry report confirms this: while AI is rapidly penetrating the travel distribution chain, the industry's actual state is one of widespread experimentation, localized applications, and room for expansion. Lack of trust, difficulties in workflow integration, and insufficient training are the main barriers to AI adoption on the business side.

This means that moving from "usable" to "trustworthy" requires more than just breakthroughs in front-end products; it demands foundational capability-building deeper in the supply chain.

Some companies are already making strides in this direction. Take DaoLV Tech, for example. Founded in Shenzhen in 2012, this company is a leading intelligence-driven global travel resource distribution service provider, covering over 200 countries and regions. With sustained growth in overseas revenue, it's one of the faster-growing companies in the industry.

DaoLV Tech aims to reconstruct the "infrastructure" of the travel supply chain—providing technology-driven distribution services for global hotel and airline resources to connect upstream suppliers and downstream channels more efficiently. AI plays a crucial role in this resource channel reconstruction process.

3

Let's examine what this company has accomplished.

In 2025, DaoLV Tech deepened its technological layout in AI, launching multiple AI application scenarios, including dynamic weight-based multi-source hotel optimal matching and AI-powered hotel semantic matching capabilities.

In October of the same year, DaoLV Tech introduced a new group structure, establishing "AI-First" as the core of its global strategy and integrating AI capabilities into multiple links of the hotel supply chain. This positioned it as a differentiated participant and collaborator in the industry's AI transformation.

More significantly, in April this year, DaoLV Tech partnered with Tsinghua University by signing a five-year strategic cooperation agreement with Tsinghua x-lab. Under this agreement, the two sides will collaborate deeply across three dimensions: tourism technology innovation, industry-academia-research project incubation, and innovative talent cultivation. They also announced plans to jointly launch a global tourism technology innovation competition, with the first stop in Europe.

The significance of this partnership extends far beyond "corporate sponsorship of a university."

Professor He Ping, Associate Dean of Tsinghua University's School of Economics and Management, summarized DaoLV Tech's unique value as a partner in nine words: "real scenarios, real problems, real implementation."

He explained that the biggest bottleneck for AI implementation in the travel industry is the long supply chain and inconsistent data standards, making it difficult to provide high-quality data inputs for AI. Tsinghua can leverage its academic strengths to collaborate with industry players in establishing data standards and breaking down data barriers through key technologies like privacy computing. However, all of this requires real industry problems as targets.

DaoLV Tech's role is to transform the most authentic pain points in the global travel supply chain directly into innovation projects, ensuring that university research remains grounded and focused.

Professor He Ping summarized this model as a closed loop: "problems come from industry, talent goes to industry, and outcomes are applied in industry." The industry side first raises real problems, the university allocates resources accordingly, academic and industry mentors provide joint guidance throughout the process, and industrial implementation effectiveness serves as the core evaluation criterion for outcomes.

Currently, "travel + AI" projects incubated at Tsinghua x-lab cover multiple directions, including intelligent distribution, accommodation experience innovation, and sustainable travel. Both sides also plan to offer specialized courses and customized training camps in tourism technology.

The significance of this exploration lies not in a single company gaining a competitive edge but in providing a collaborative paradigm that the entire industry can reference. When industry-academia-research forces truly integrate, the entire cultural and tourism technology sector will benefit.

4

After learning about the innovative attempts and explorations of various travel companies around the new technology of AI, should we consider the "Doubao incident" a sign that relying on AI for all travel matters is truly out of reach and not worth anticipating for mass tourists?

The answer is: we can be disappointed with the present, but we shouldn't doubt the future.

Any disruptive technology must undergo an evolutionary process from "usable" to "user-friendly" and finally to "trustworthy" in its early stages. Today's AI travel assistants are like mobile maps in 2010—you might end up in a ditch after entering an address, but no one would deny that mobile maps ultimately revolutionized human travel.

Moreover, this transformation will only accelerate. According to a survey by global market intelligence and research firm The Business Research Company, the machine learning market in the tourism industry is expected to grow from $3.78 billion in 2025 to $4.45 billion in 2026, with a compound annual growth rate of 17.7%.

The tide of capital and technology has already begun, and industrial transformation is irreversible.

Whether it's heavy investments by platforms like Ctrip and Tongcheng in front-end experience, Fliggy's continuous exploration of multi-agent architectures, Mafengwo's Agent evolution in content ecology, or backend technological foundations built by B-end resource platforms like DaoLV Tech, along with its industry-academia-research experiments with Tsinghua University—these forces from different directions are all driving toward the same goal:

Transforming AI from a mere "advisor" that offers suggestions into a trustworthy "butler" capable of managing entire travel itineraries.

The ultimate goal of AI in travel is not to replace human curiosity and exploration of distant places. Instead, it aims to free us from exhausting price comparisons, tedious planning, and endless confirmation calls, then return the essence of travel—the freedom to set off on a whim and the enjoyment of scenery and cuisine—back into our hands.

We should believe that day may arrive sooner than many people imagine.

Images sourced from Shetuwang and online screenshots