When ChatGPT launched publicly in late 2022, it triggered immediate discussion across every industry about what large language models would change and how fast. This paper was one of the earliest academic responses from the hospitality and tourism field — published just months after the public launch — and it still holds up as one of the clearest maps of the opportunity and the risk.
The authors systematically identify concrete use cases across the three major stakeholder groups who interact with tourism: tourists and travelers, tourism businesses (hotels, airlines, tour operators), and destination management organizations.
For travelers, the most immediately useful applications are itinerary planning and customer service. ChatGPT-style tools can build personalized travel itineraries based on conversational input — integrating preferences, budget, timing, and interests in a way that no traditional search interface can match. For routine customer service (reservation queries, policy explanations, complaint triage), AI chatbots can handle a high proportion of incoming contacts without human intervention, reducing wait times and freeing staff for complex issues that genuinely require judgment.
For hotels and tourism businesses, the paper identifies back-office productivity as the fastest-win category. AI tools can generate property descriptions, social media posts, email campaign drafts, menu descriptions, and review responses at a fraction of the time cost of manual writing. Revenue management and dynamic pricing represent a higher-complexity opportunity where the technology is developing rapidly.
For destination management organizations, the paper highlights content generation and visitor information as high-value use cases — AI that can answer visitor queries 24/7 in multiple languages without a corresponding increase in staff costs represents a genuine step-change in what smaller destinations can offer.
The risk assessment is where the paper earns its longevity. The authors don't minimize the problems. Accuracy is the most immediate: language models generate plausible-sounding content that is sometimes factually wrong, and in a service industry where trust is foundational, that's a serious risk. Hotels that have deployed AI-generated content without human review have published incorrect pricing, wrong amenity descriptions, and in some cases outright fabricated attractions nearby. The paper recommends treating AI output as a first draft that always requires human verification — not a finished product.
Workforce implications are handled with more nuance than most early commentary on the topic. The paper distinguishes between tasks (many of which can be automated) and jobs (most of which are more complex than any single task). The realistic near-term scenario is that AI tools change what hospitality workers do rather than eliminating their roles — but this transition requires active management, not passive observation. Organizations that build AI fluency into hiring and training now will be better positioned than those that react when the change has already happened.
The most useful practical takeaway for operators: start with use cases where accuracy errors are low-stakes and easily caught — social content, internal communications, first-draft materials — and build your team's confidence and review processes before moving to guest-facing or revenue-critical applications.