A fantastic day in Los Angeles commences with a leisurely walk along the Venice Beach boardwalk, followed by a delightful ride on the Ferris wheel in nearby Santa Monica. Next on the agenda is a visit to the Getty Museum, located approximately nine miles away by car. Subsequently, a trip to Beverly Hills is in order, followed by a visit to Hollywood to explore the iconic Walk of Fame. A hike in Griffith Park offers a serene retreat before indulging in delectable dim sum in Chinatown. The day concludes with a visit to downtown, where catching an evening show at the Walt Disney Concert Hall is a perfect finale.
This meticulously crafted itinerary was customized for me by Microsoft Copilot, an AI assistant, in response to my request for a tailored one-day exploration plan. While the suggested schedule may seem ideal for a visitor, it may not align with the preferences of a local resident like myself. The bustling nature of traversing through Los Angeles, a sprawling and traffic-laden city, from one landmark to another may not necessarily epitomize a perfect day for everyone.
AI companies often promote their chatbots as versatile tools for travel planning, offering assistance in selecting destinations, comparing flight prices, and discovering both popular and off-the-beaten-path attractions. However, despite the wealth of travel information available online, chatbots still struggle to generate truly personalized and comprehensive travel itineraries.
In a study conducted by researchers from Fudan University, Ohio State University, Penn State, and Meta, various chatbots were tested on their ability to create travel plans based on specific criteria. The results were underwhelming, with even the most advanced model, OpenAI’s GPT-4, successfully fulfilling only a small fraction of the queries.
These AI chatbots often fall short due to reasoning errors and inaccuracies in their responses. While they aim to supplement human decision-making processes, they currently lack the finesse and attention to detail required for effective travel planning. Despite these limitations, AI companies like Microsoft and Google emphasize the time-saving benefits and information aggregation capabilities of their tools in travel planning.
Looking ahead, the potential for AI to revolutionize travel planning remains promising. Integrating features like flight booking and map navigation into chatbots, as seen in Google’s Gemini chatbot, hints at a future where AI can facilitate seamless trip planning and execution. As AI technology continues to evolve, there is optimism that these tools will eventually offer more sophisticated and tailored travel experiences.
In the realm of AI development, the focus is shifting towards practical applications beyond consumer-facing tasks, such as machine learning in scientific research and healthcare. While AI’s impact on travel planning may still be evolving, its potential to transform industries like healthcare and science is increasingly apparent. As we navigate the evolving landscape of AI capabilities, the allure of effortless travel planning with chatbots serves as a reminder of the ongoing advancements in artificial intelligence.
Caroline Mimbs Nyce is a staff writer at The Atlantic.