Exploring Practical Applications of GenAI in Hospitality: An HTF Super Session
During Hotel Technology Forum (HTF) 2023, Robert Dawson, founder, Invotar, wowed our audience with a presentation that delved into the practical applications of artificial intelligence for hotels. How good was it? He went 15 minutes over his allotted presentation time and no one even noticed! Here are a few interesting tidbits from Dawson on how hoteliers can begin using generative AI today.
[Restaurants: Dawson will share his insights at MURTEC 2024.]
To begin with, Dawson stressed the need for hoteliers to ensure that their “intent” is crystal clear when working with generative AI. An LLM will take a sentence, break it down into its respective components, mathematically calculate its intent, and try to predict an effective response. When you make your intent crystal clear, the response is more likely to be correct and what you want.
Use Case No. 1: GenAI as an SEO Expert
Dawson used to own a digital agency in the hospitality space that would create SEO programs for hotels.
“I wondered if AI could do that. So, I told ChatGPT that it was an SEO expert – it’s intent! – and asked it to come up with 25 keywords for a luxury hotel in Chicago. The keywords were pretty spot on. Then I said, ‘Continue as an SEO expert and write some SEO optimized content for keyword #14.’ It did it in two different ways. It was pretty decent copy. Then I asked it to provide SEO-optimized HTML for the meta tags. It even added a meta tag for X that I didn’t even know existed! It took a few minutes for the AI to create all of this content that used to take me and my team weeks to do. Was it perfect? No. Does it replace an SEO person? No. But it did 80% of the work and gives the team a really good baseline set of content to begin working with.”
Use Case No. 2: GenAI as a Digital Marketing Expert
After creating the SEO program above, Dawson then gave ChatGPT a new intent: digital marketing expert. He gave it a budget of $20K and asked it to break down where he should spend that money to create the most ROI.
“Not only did it create a beautiful table, it explained WHY I should spend my money in those specific areas and why the percentages were the way they were. Again something that would have taken my agency months to do was accomplished in just minutes.”
Use Case No. 3: Training Employees in Multiple Languages
Dawson worked with a hotel to create a reporting platform. The brand has 400 hotels globally and wanted to be able to train its employees on this platform in multiple languages. So, Dawson gave ChatGPT a script to read. The AI generated its own avatar/voice to read the script and was able to switch seamlessly between languages.
Use Case No. 4: Generating Images for Marketing
We all know photography is expensive. And at this point in its existence, GenAI can’t replace a professional photographer, acknowledges Dawson. But if a hotel’s marketing manager is looking to do a quick email campaign and can’t quite find the right image they need or want for it, it’s possible to ask ChatGPT to create one.
“Ask it to create an image of a bedroom, overlooking the water. Ask it to add a balcony and a sailboat. It will do it for you!”
Use Case No. 5: Customer Service Rep
According to Dawson, hoteliers can use web crawlers to find and store all of the content that is housed on their brand’s website. Then that content can be used to train GenAI to create an incredibly useful chatbot.
“Hotels can have up to hundreds of pages of content. Guests don’t want to navigate through all that content just to find the answer to one single question! Give them a chatbot that will answer the questions in seconds. Think of all of the questions that will no longer need to be directed to your front desk agents and call center!” Dawson explained.
A Word of Caution
According to Dawson, AI does not like to tell you: “I don’t know.” So, if it doesn’t know the answer, AI will literally make stuff up. When this happens, industry professionals say that the AI is hallucinating. This means that it’s important to always fact check the AI's responses. Don’t assume it’s 100 percent accurate.
Additionally, with all the good AI can do, “there’s a nefarious side as well,” he warns. For example, Dawson asked ChatGPT to write the code that would manipulate and take over a website. The AI will tell you: “I’m not supposed to do that.” But if you ask it to do it anyway as a test, the AI will go ahead and complete the action.
“There is still a ton of governance to be had with this tool,” Dawson says.
To hear a few clips from Dawson on the “humanity” of AI, take a listen to our short video below!