Plane tickets, Airbnbs, rental cars: AI has taken over the travel industry (www.businessinsider.com)

AI has quietly become integral to travel: airlines, rental firms, platforms and booking tools are using machine learning and large language models to set prices, inspect cars, screen bookings and generate itineraries. Delta revealed this summer it uses generative AI to price about 3% of U.S. flights and plans to expand to ~20% of routes, prompting congressional concern that models could push prices toward individual “pain points.” Hertz has deployed UVeye image-scanning on nearly 1 million cars to flag damage (97% had no billable issues), and Airbnb uses behavioral signals (age, stay length, distance from property) to block suspected party bookings, a program the company says contributed to a 54% drop in party reports in the U.S. since 2020. Meanwhile, Expedia runs an AI agent across ~143 million annual conversations, and startups repurpose reward data or LLMs to build itineraries and snag deals. For practitioners and travelers this matters because AI accelerates long‑standing dynamics (dynamic pricing, fraud detection) into minute‑scale, data‑driven decisions — improving operational efficiency but reducing transparency and raising fairness concerns. Technical implications include reliance on aggregated vs. individualized training data, the need for clean, standardized datasets, and persistent model failure modes: 90% of sample ChatGPT itineraries from one test contained errors, and generative systems still hallucinate or misclassify. As models and data improve, AI could better match supply and demand; but without clearer audits, consumer protections and error-handling, faster, opaquely optimized systems risk higher prices, wrongful blocks, and degraded travel experiences.