Can you book flights and hotels using AI? The answer is that while AI chatbots have the potential to assist with booking travel accommodations, they are currently unreliable and require additional research from the user.
To illustrate this, let’s compare the experiences of using two AI chatbots – Bard and ChatGPT – to book flights and hotels. In my recent scenario, I spent hours researching flights and accommodation for a three-week trip to Japan. I decided to test the capabilities of Bard and ChatGPT to see if they could provide helpful suggestions.
Bard, surprisingly, proved to be quite adept at finding flights. When I requested flights from Melbourne to Tokyo on a specific day, Bard presented options from reputable carriers such as Qantas and Japan Airlines. It was able to refine the results based on criteria like the cheapest direct flight with seat selection, at least 15 kilograms of luggage, and a meal. Bard even found a cheaper Air Asia flight from Melbourne to Osaka compared to the one I had already booked to Tokyo. The AI was even able to determine seat width, pitch, and recline angle for the Air Asia flight, providing valuable information for potential passengers. Although Bard couldn’t provide a direct booking link, I verified that the prices and details matched with the airline’s website.
On the other hand, ChatGPT, despite its new Kayak travel agent plugin, failed to deliver accurate results. It suggested a 29-hour flight with layovers in Atlanta and Detroit, which was three times longer than a direct flight. Despite the availability of direct flights, ChatGPT insisted that there were none. It’s worth noting that ChatGPT is more focused on the U.S. market, so results may vary depending on the region.
Related Post
When it came to hotel recommendations, the Kayak plugin in ChatGPT emerged as the winner by default. Prompted to find an affordable double room in Shibuya with a review score above 7, it suggested the Shinagawa Prince Hotel for $155 a night and provided a direct booking link. However, it turned out that the hotel was an hour’s walk from Shibuya, and none of the other options were located in Shibuya either. Bard’s hotel suggestions fared even worse, offering a non-existent Hotel Gracery Shibuya at $120 a night. When I attempted to book the Shibuya Excel Hotel, Bard quoted a price of $100 per night, but the actual cost turned out to be $220 a night. Bard’s fictional scenario continued when it provided a non-existent transcript of our conversation, including a booking at the non-existent Hotel Gracery Shibuya.
These experiences highlight that while AI chatbots have potential, they are still far from reliable when it comes to booking flights and hotels. Users would still need to conduct their own research and cross-reference the information provided by AI chatbots. It’s clear that AI assistants have the potential to revolutionize travel booking, but there is still progress to be made, and the reliability of their suggestions needs improvement.
In other AI news, Toyota has unveiled generative AI tools for designers to create new car concepts. Designers can input a rough sketch and text prompts to describe the desired characteristics, and the AI will transform it into a finished design. Vimeo is introducing AI script generation to its video editing tools, allowing users to type in subject matter, tone, and length to generate a script. China Science Daily claims that Baidu’s Ernie 3.5 outperformed OpenAI’s GPT 3.5 in qualification tests and can beat GPT-4 in Chinese language tests.
Booking.com has launched an AI Trip Planner for select Genius-level app users to help with itinerary planning and accommodation booking. While Google’s Bard has seen significant growth in visits, it is still less popular than ChatGPT, which saw over a billion visits in May. Google is also leveraging Alpha-Go AI techniques for its latest model, Gemini, to surpass the capabilities of GPT-4. The GPT Portfolio, which entrusts trading decisions to ChatGPT, has garnered $27.2 million in copy trading, but its returns have been underwhelming compared to the S&P 500.
In the realm of crypto plugins for ChatGPT, several options have emerged, including SignalPlus for NFT analysis, CheckTheChain for wallet transactions, CryptoPulse for crypto news analysis, and Smarter Contracts for analyzing token or protocol smart contracts. However, these plugins are still finding their footing, and their utility may not yet surpass dedicated websites in terms of accuracy.
Overall, while AI chatbots show promise, especially in the travel and crypto industries, there is still room for improvement. Users should approach AI chatbots with skepticism and conduct their own research to ensure accuracy and reliability.
