Guest Satisfaction: Can Artificial Intelligence Be Beneficial to Hotel Guests?
By Gustaaf Schrils Chief Information Officer, White Lodging | January 05, 2020
In November, "Terminator: Dark Fate" hit theaters. This franchise is all about hybrid human cyborgs and time travel, all to prevent the genesis of machine learning and artificial intelligence. Perhaps, fiction and real life are in parallel with the emergence of these scientific initiatives being implemented around the world today. You can now feel safe shopping in a mall in Dubai that enlists robot police officers in its security force. The city also has plans to start using automated patrol cars, drones and robots to monitor parking, issue fines, predict crime, find abandoned vehicles, and, analyze traffic patterns to predict where bottlenecks might pop up[MB1] .
The perfect storm is hitting us now. That storm being the convergence of Artificial Learning (AI), Machine Learning (ML) and human beings. The public have become accustom to marketers tracking cookies on web searches, Siri and other voice assistants, and even watching Watson win a Jeopardy match. The advent of machines in our day-to-day lives is here. What we need to remember, and continue to employ, is the human contribution to all of this. Unlike the Terminator saga where machines become masters of their own destiny, in our world that is not the case. Machines are (still) a tool and human intelligence is the force on the lever of innovation.
The hospitality industry is ripe for these leaps in processing data to better serve our guests. But first, let us distinguish between applications. It is easy, and somewhat mistaken, to think that AI and ML being used to make a reservation recommendation based on past behavior of one guest is a good example. Working on such a small dataset is not really where AI shines. Instead, analyzing a vast database of complete and incomplete reservation transactions combined with behavioral indicators that will produce demand and rate forecasts with guidelines for the pace of booking is a much better example.
In hospitality we could look at two general areas in which AI/ML could have major impacts. First, would be the guest life cycle. The guest life cycle extends from the moment a guest thinks about travel and hotel stays to after they have checked out and returned home to share their experience with friends and family.
Analyzing data in regards to yields from advertising in various markets, via different methods and periods such as seasonal or time of day, could design the marketing content and efforts that any one potential guest might encounter. From that point, based on their status within the booking engine they might be prompted, or lead, to lean one way or another based upon financial returns and predicted availability. Once booked and in route, influences such as previous stay data, estimated time of arrival, weather, and behavior trends of multitudes of other guests with the same demographics and economic status can drive delivery of pre-stay content. While these examples appear to be individually focused, in fact they are designed to drive efficiency and profitability across thousands of hotel rooms at any given moment.
Upon check in, factors such as length of stay, forecasted occupancy, financial means and past behavior could provide relevant upsell opportunities from the hotel to the guest. During the guest stay, researching local events, group itineraries, close restaurants and their specials, all can deliver focused excursions for guests to experience. Chatbots, and voice assistants like Alexa that employ Natural Language Processing can expedite the delivery of requested information while guests are contemplating next steps. Even the potential of extending check-out a few hours after analyzing weather and flight schedule delays can deliver on both guest satisfaction and additional revenues.
A second area within this industry that AI/ML can be of benefit would be operations and profitability. Successful AI/ML implementations will be rich in Deep Learning; meaning that these engines will learn from mistakes and improve upon the algorithms that humans have programmed into them. One such area could be labor management. Taking into account a myriad of factors such as future occupancy with guest count, calendaring events in the local area, tracking current employee trends, analyzing unemployment and labor pool over seasons, all could develop more efficient labor forecasts and schedules.
Efforts could be made to extend the useful life of the physical asset of the building itself. Preventive maintenance schedules, occupancy forecasts, seasonality, age of systems, manufacturer recommended practices, available labor, and more, can all be analyzed to design programs to protect the physical asset of the building as well as assets within the building.
These examples are just a few of the applications that can be dreamed up for AI. AI can be a robust platform that will offer many benefits but will really only succeed by the integration of the human mind and imagination creating the algorithms that drive these machines to do what they do. Don't discount the importance of the human factor. There is a very old computer programming acronym, GIGO, which stands for "Garbage In, Garbage Out." If we only ask a machine to deliver narrowly focused, irrelevant data that is all we will get.
The human component of this hybrid combination is to secure voluminous, pristine, linked data that with the right questions, the right algorithms, machines can then detect patterns. The patterns in and of themselves may be ground breaking or could be completely irrelevant. Machines do not (yet) have the capability to judge the value of the patterns and responses they spit out. Of the six levels of cognition (knowledge, comprehension, application, analysis, synthesis and evaluation), machines, at best, can function at the analysis level – which is critical thinking; the ability to break down copious amounts of data into relative nuggets. Humans still carry the most important functions in this marriage, that being synthesis (creative thinking) and evaluation (judgement).
Designing a worthy AI application requires data and the right questions, but also must allow for the mistakes that will most certainly happen and make corrections. Data may be inaccurate or even missing; meaning that very relevant data that if included in the process could improve the deliverable greatly, may simply be overlooked. [MB2] [MB3] The questions we ask of the machine may not be presented in a manner that will yield relevance. [MB4] And, machines will, and believe it or not must, make mistakes. ML, and Deep Learning, are dependent upon recognizing mistakes and then learning from them and correcting them to improve. It would be to our detriment to configure an AI engine, believe that we have accurately completed what we planned, and then sit back and expect actionable data consistently.
The Terminator movies take this mistake to the extreme. In the movies, humans have created AI and ML and have provided the data and question sets that the machines then come to the conclusion that humans are actually detrimental to the well-being and health of the world in which we live. The lesson here is that humans must create the questions, the algorithms that will drive the engines to deliver actionable, relevant, valuable data for us to apply to our benefit, and, continually evaluate the output.
This lesson should not be overlooked in our efforts here in the real world. Perhaps the best integration of human input leveraged with AI/ML would be to predict the times in which we need to make decisions and then present us with the relevant data and options from which to make the best decisions. In hospitality, on the individual scale, this would result in applications that, for example, send a dinner coupon to a hotel guest in their room at 6 PM to a local restaurant that provides a favorite cuisine accompanied by a select bottle of wine.
On a grander scale, for example, the machine could recognize that the 2020 Democratic National Convention will be held in Milwaukee in July, determine room count within 15 miles of the event, screen registered Democrats that are politically active, between the ages of 25-50, that travel 3-5 times per year, that have an income of over $75,000, and then pick the hotels and rates to offer to the highest percentage of this data set likely to book rooms for the event within two weeks of the announcement.
Artificial Intelligence, Machine Learning and Deep Learning are already here. There are wonderful applications just waiting to be dreamed up. The successful applications will be the ones that deliver personalized service based on the best decisions for the overall business. Our guests deserve the highest levels of service without feeling targeted or manipulated. They want to be identified as unique and special, not feeling that they are a number in a jumble of spreadsheets. It is the human factor that needs to craft the algorithms that makes these feelings possible. And, as far as this writer knows for now, humans are still the most critical part of this associated relationship.
- Chatbots – "a computer program designed to simulate conversation with human users, especially over the Internet"
- Artificial Intelligence (AI) – "the capability of a machine to imitate intelligent human behavior"
- Machine Learning (ML) – "the capability of a machine to improve its own performance"
- Deep Learning – "a type of ML and AI;" "an important element of data science, which includes statistics and predictive modeling"
- Automation – the ability of a system to execute variable repetitive tasks without human intervention
- Natural Language Processing (NLP) – "the ability of a computer program to understand human language as it is spoken"
HotelExecutive retains the copyright to the articles published in the Hotel Business Review. Articles cannot be republished without prior written consent by HotelExecutive.