By Rob Landon, VP of Engineering, Knowland
The technology industry has been buzzing since ChatGPT broke onto the scene in late 2022. As so many of us have done, I gave it a try. I bombarded it with all kinds of questions and requests. My inner Star Trek nerd raised an eyebrow. It felt like a significant step toward that science fiction future: a computer with access to the collective knowledge of humanity you can converse with like it is another person. That’s exciting stuff.
Once the initial wonderment wore off, I started thinking about it both within the hospitality industry and in general. How far along is it, and what can we do with it in our business applications?
I asked ChatGPT, “What is ChatGPT?” It replied:
I’m ChatGPT, a large language model trained by OpenAI. I use machine-learning techniques to generate human-like responses to natural language input. My purpose is to assist users in answering questions, completing tasks, and engaging in conversation. I have been trained on a large dataset of diverse text, which allows me to generate responses that are both relevant and informative.
It responds in first person? Of course, it does. What does it really bring to the table, though? It is undoubtedly an extension of existing efforts and successes in the AI/ML (Artificial Intelligence/Machine Learning) realm. What we all want to know is how much of an extension it is. And while it’s a very interesting trend, is it a giant leap?
In the realm of search, the big players have been using AI/ML to feed users more relevant results to their searches for some time. The significant jump for search using ChatGPT is how this interaction is accomplished. Instead of typing keywords and getting a list of possible results sorted by relevance, the user can have an actual conversation with the search engine. The continuity of the conversation allows for the refinement of this search and future searches. It seems clear that this will likely be the way we all will soon interact with search engines.
Using ChatGPT within typical business applications
For those of us building SaaS applications, many use proprietary data in part or whole. The bad news is that ChatGPT is, at its best, in its infancy in the proprietary data realm. ChatGPT is hosted on cloud-based servers owned and managed by the entity that developed and trained it, OpenAI. Its data sources are public and meant to be used by the general public. This presents real problems for proprietary use on several fronts, namely security, privacy, and contracts. Yes, the underlying technology, GPT-3, is available to third parties and can be hosted to allow contained, proprietary use, but it requires a lot of work, horsepower, and a large dataset for training.
Application vendors that are not constrained by the inclusion of proprietary data, should be evaluating ChatGPT now. Even if their data is currently not publicly available, but it fits into their business models that allow it, they can make that data available to ChatGPT through APIs. Everything is still early, even for these types of applications, but the barriers are much, much lower.
In the hospitality industry, several applications come to mind: virtual concierges, informational kiosks, virtual front desk assistants, etc. Yes, these applications exist already, but adding natural language as an interface will improve the experience of these dramatically.
For the rest of us with proprietary data, it’s even more early. On March 1, 2023, OpenAI released developer APIs for its popular ChatGPT and Whisper AI models. This is the type of capability necessary to head in the right direction. However, data privacy and security issues still loom over the distribution, storage, and use of proprietary data.
Make no mistake, though, the human-computer natural language interfaces of science fiction are coming and will be here soon. Even if it’s early, we should speculate about these possibilities and explore where they will impact our applications in the not-too-distant future.
Natural language impact on the workflow of applications
Most applications still use filters to dig through their data. Most of us are very familiar with the plethora of applications that use a combination of entry boxes, radio buttons, drop downs, etc., to filter large datasets into something useful. The more advanced applications use AI/ML to assist in this process by scoring data, forecasting, and pattern recognition. Even when aided by AI/ML, the process can generate too many or too few results.
However, natural language interfaces will completely revolutionize the workflow of applications. The interactive conversations of all users will make applications flow in ways that are more natural to the task at hand. Instead of working through coded workflows and tools, AI will be able to create the flows and anticipate the subsequent desired actions. We already see this in many applications, but they are not driven and displayed with natural language expectations.
Using AI in the hospitality industry
The hospitality industry has embraced AI for years in areas like revenue management by forecasting demand and generating future pricing at segment levels. As AI tools become easier to use/train/integrate, forecasting will become more and more useful. Interacting with, generating, and re-generating these forecasts with a natural language tool like ChatGPT will remove a lot of tedious work currently being done by a human and allow them to work on other, more important things.
Often, a large amount of data is used to make decisions, but people need help to make those decisions reliably, in volume, and quickly. As mentioned earlier, inputting into searches and filtering the results is typical. When we can replace that interaction with a natural language interface, users should have an infinitely easier time sorting through the data. Many hospitality applications use this pattern and leverage various methods to provide relevant results or outcomes. Imagine having a conversation with your Property Management System, Revenue Management System, Sales and Catering System, or related applications as you try to extract data, create report templates or look for future business.
Not ready for Prime Time…Yet
ChatGPT (and its competitors) are here to stay as a technology concept. It’s too early to tell who the winners will be, but the concept of talking naturally with a computer is coming. Indeed, it will revolutionize the way we build and interact with applications.
We’ve embraced AI/ML in several areas in the hospitality industry for years. ChatGPT is a very real, subsequent, natural step. Sorting through data can be tedious, and the current workflows can be as well. Having a conversation to get through this data will make the entire process easier and more efficient.
It’s still very early, but this technology will hit publicly available applications first. Applications dealing with proprietary data, either the source input, processed output, or both, won’t be too far behind once we’ve figured out legal/contractual, privacy, and security concerns.