Day 19 of 30 Day AI Challenge – How AI can supercharge our understanding of Hotel Guest Sentiment

For those just joining us, every day during this challenge, I’m going to try and do something different or better, using currently available Artificial Intelligence tools.

Let’s take on today’s challenge: to use the power of AI to analyze heaps of data and uncover the heart of the matter. To test this, I had ChatGPT act as a hotel review analyst. I fed it plain text reviews from public forums and asked it to summarize the key insights.

ChatGPT as your personal hotel review analyst:

Here’s an example of the initial prompt and hotel review input:

Here’s what ChatGPT produced:

I probed further to see if it could pinpoint what sort of guests were coming in and what they were most interested in:

I also asked for some basic sentiment analysis:

Although the test above involved a small sample size of fewer than ten reviews, the potential for this system is far greater. Imagine feeding the AI ALL the available reviews from various sources (TripAdvisor, OTAs, Hotel Website, Guest Survey Responses**) and conversing with it naturally to obtain answers. This surpasses current options, as even the most advanced tools and services require significant effort to extract insights and can have complex interfaces, leading to a learning curve. However, integrating GPT4 (or a similar large language model) into these tools, with the ability to generate visualizations and work with closed data sets, would revolutionize guest sentiment analysis and allow us to identify the underlying causes of specific issues quickly and intuitively.

How ChatGPT handles more complex data sets

I also wanted to push the AI to its limits and see how it handled (for want of a better term), essentially a structured data dump. To test this, I grabbed a small portion of publicly available hotel review datasets (source), removed a few columns, and handed it over to ChatGPT. I asked it to dive into the data and see what it could uncover – any insights it could provide.

This was truly impressive and highlights the incredible potential of an AI-driven review and sentiment analysis tool. No doubt, software providers are rushing to leverage these newfound capabilities. The implications are enormous – this could be a game-changer not only for hotels, airlines, and hospitals but any industry that relies on customer feedback and reviews to stay ahead of the game.

Keen to see who moves quickest, and what they come out with!

Till tomorrow…

**It goes without saying that you need to be thoughtful about exactly what you feed the system, especially large models that may not offer the level of data protection or privacy needed for certain kinds of sensitive or proprietary data. Just as you wouldn’t put internal data publicly on Facebook or Twitter, you shouldn’t feed it into large public LLMs (generative AI models) either. As the tech and tools develop, there will be ways to ring fence data as required by company policy and local law – till then, please be cautious about what you use off-the-shelf AI tools for!