With the advent of more accessible AI technologies, there is a new interest in the practical use of AI in the travel industry says Upali Kohomban PHD, VP technology at CodeGen
Guest Post: Leveraging AI in the travel industry: Beyond the hype
Artificial intelligence (AI) and machine learning (ML) have been buzzwords in the business world for quite some time, but it has only been in recent years that these technologies have started to make a significant impact on the consumer industry. With the advent of more accessible AI technologies, such as OpenAI's ChatGPT, there is a new interest in the practical use of AI in the travel industry.
While the availability of machine learning and AI libraries such as TensorFlow, PyTorch, Scikit-learn, Keras, and Theano have enabled the scientific community to develop complex AI algorithms, these technologies required highly specialized skills and a deep understanding of relevant theories. However, new technologies, such as ChatGPT and its counterparts, have made it easier for non-experts to leverage AI for practical use in the industry.
In the past, periods known as "AI Winters" have occurred when AI was hyped up by some new invention, followed by a failure to deliver the market expectations This led to a decline in interest and funding for AI research. However, in my opinion as someone who has been using AI in travel applications for more than a decade, the current interest in AI in consumer industries, including travel, is not likely to lead to another AI Winter, as long as the hype is looked past.
There are several areas in the travel industry where AI can be used to improve customer experience, such as:
- sentiment analysis
- recommendation and personalization
- revenue management
- and customer relationship management.
Although GPT is a groundbreaking technology in terms of its accuracy and user-friendliness, its conversational AI capabilities have certain limitations that may not be able to address all the information requirements of the travel industry. In order to meet these requirements, the industry will still need to employ different ML technologies. One example is collaborative filtering, which is a well-established and proven method for implementing recommendation systems. In simple terms, collaborative filtering assumes that if you have the same preferences for certain things as a group of people, then you are likely to prefer other things that the same group of people like. Many businesses, such as Amazon, employ similar technologies for their recommendation systems, which have proven to be very effective. This is an example of an ML application that goes beyond the scope of GPT.
What will eventually be interesting to consumer applications is how the industry will build upon GPT’s natural language processing abilities, in order to enrich another level of business intelligence, by pre-processing and analyzing data with the newly available power and then using ML and AI on the results to add business intelligence.
For instance, the user reviews can be sentimentally analyzed in order to:
- provide better recommendations
- Directly answer users’ questions
- Highlight the salient points about an entity
- Provide feedback to the entity vendors about user reception.
Or, if we consider recommendation, a truly intelligent system will be able to:
- Consider the personal preferences of the user
- Remember and infer the preferences from the users’ past selections
- Optimize the recommendations for revenue by carefully balancing both look to book ratio and the profit margins and availability
- Learn from its own mistakes via how the users pick its recommendations
All these tasks go beyond the scope of natural language, and involve other areas of AI at large, such as reinforcement learning, association rule mining, and in general data mining and statistics. It will be this value addition that will actually determine whether the end user application is revolutionary.
At CodeGen we're currently introducing the latest machine learning models across our product stack as part of our AI First strategy which will help our clients save time on everyday tasks like contract loading, hotel/room mapping, NLP search and product recommendation.
To find out more about CodeGen’s products and services come and meet us at ATM Dubai on stand TT1160.