Machine Intelligence in the Travel & Transportation Industry
Chute releases a travel-marketing magazine called Sightseer each quarter. A version of this article will appear in the Q3 edition.
TL;DR:
Strive to find the right balance between human and automated interaction with customers. Convenience is equally as important as empathy.
Focus on process improvement prior to process automation. Automating bad processes increases the velocity at which bad things happen.
A History of Technology
Machine intelligence has been changing the way we travel for generations. In 1946, American Airlines released the first automated booking system. By 1976, United Airlines deployed their Computerized Reservations System to travel agents. In the 1990’s, Lonely Planet’s first website, Expedia, and Priceline were born. Throughout the early 2000’s, Global Distribution Systems and the Internet enabled Kayak and Hotwire to innovate the OTA model with aggregation. Today, nearly every airline has an “intelligent” (or at least dynamic) application; major hotel chains are deploying automated concierge advisors; the TSA utilizes myriad scanning and visual recognition security technologies; travelers’ preferences and digital behavior are being used to provide tailored offerings, and they can even purchase luggage that follows them between terminals at the airport (see Travelmate).
Throughout the past few decades however, there has not been consistent tenacity for invention. The progress achieved during 1960’s machine translation efforts and the proliferation of “expert systems” in the 1980’s each gave way to their own research funding collapses. These decades of innovation dormancy are commonly referred to as “AI winters.” While some believe we are prone for another, many claim certainty that the AI winters of past generations are now completely in hindsight. Whichever opinion you may hold, the most recent uptick can be traced to several key technology and infrastructure enablers:
Centralized and Decentralized Computing
Advanced data manipulation techniques require significant compute power. Companies like Amazon Web Services and Google’s Cloud division centralize servers (and GPU clusters) to allow software applications to run off-premise or “on the cloud” and stream to devices. In parallel, within the palm of our hands, the hardware used in mobile devices today is remarkably advanced — enabling substantial computation to occur on the device itself, or “at the edge”.
Mobile “Appification” and Inter-App Services
Mobile is the largest economic platform in the history of humankind. As such, millions of apps have been created. End users navigate through these apps manually today and typically spend 85% of their time-on-device between only five apps. As less pervasive and niche consumer services come online as apps, it becomes increasingly vital for operating fabrics to exist between them. Button is a great example of this type of service — enabling end users to seamlessly navigate between apps to complete tasks like scheduling an Uber ride to get to an OpenTable reservation.
Natural Language Processing (NLP)
Roughly 80% of the information on the internet (and within the enterprise) is considered unstructured data — information that is neither pre-defined nor organized. Unstructured information is typically text-heavy and contains data such as dates, numbers, and facts. Advancements in the field of NLP help make sense of this data by interpreting meaning, intent, and sentiment — enabling companies to have better insight into how an audience feels about a product or service.
Computer Vision
The latest techniques in deploying algorithms to learn visual cues has opened up avenues to automate workflows and business processes involving tagging and/or interpreting images. Chute has built its own visual recognition engine, called IRIS. It can filter and tag images by scene type, logos, and objects. Captricity is another great example — processing scanned documents and images within insurance workflows.
Training, Optimization, and Prediction
Machine learning algorithms improve performance through training data and feedback loops. Search Engine Optimization (SEO) is an example of how the likes of Google and Bing rank responses on a large scale. At scales large and small, optimization algorithms are being deployed to leverage past online or digital behavior to recommend future activities, continuously calibrating and tailoring suggestions over time.
Voice Recognition
Voice Recognition technologies are now yielding around a 5% error rate, which is at parity with human abilities. Voice activated agents like Apple’s Siri and Amazon’s Alexa are opening the aperture on ways we can simply ask to activate the myriad services on our mobile devices. Think of Interactive Voice Response (IVR) on customer service calls that actually works.
Developer-friendly services have been built by leveraging these enabling technologies — catalyzing adoption among technologist communities and thus driving the proliferation of applications utilizing machine intelligence today. The themes below dissect the ways in which these technologies are most drastically impacting the travel experience.
Engagement
Chatbots are a unique and effective medium to connect with customers. Facebook Messenger, IBM Watson, Heroku, Amazon Alexa, and Slack (among MANY other) platforms provide frameworks for building and training chatbots. These applications may be deployed as a concierge at a hotel helping locate the gym or an airline app’s messenger alerting travelers of delays or a voice-command interface to book a ticket.
Search & Exploration
Travelers can now search in natural language to find desired and tailored experiences wherever they go. Intelligent services like WayBlazer recommend events, restaurants, parks, museums, and experiences that will most likely resonate with the traveler by combining reviews, rankings, and travel guides.
Logistics & Operations
It is now easier and faster than ever before to book across 200+ countries in dozens of languages. Intelligent transportation systems — from human-less airport railways to breakthrough autonomous vehicle technology — is changing how travelers get from A to B. Further, dynamic pricing algorithms are shifting the way we spend money based on real-time supply and demand throughout the travel experience — during both booking and transport.
What might the future hold?
Soon enough, simply booking a flight ticket will trigger a human-less ride to be scheduled. Truly personalized restaurant and activity recommendations will be immediately and automatically sent with one-click booking features through a messenger application. Trip planning will take a fraction of the time it does today. Even after reaching the airport, a traveler will be guided to their gate with shopping, lounging, and dining suggestions along the route. The airport will be more than a means to get elsewhere; it will be a place to be in itself.
Hospitality staff will be notified by social monitoring systems when influencers arrive at the restaurant bar or check-in to a hotel, enabling the service team to proactively create an incredible experience for them — perhaps with a drink or dessert on the house. Airlines may even target specific loyal and influential fliers troubled with delays to compensate in unique ways.
Like autopilot in an aircraft, Machine Intelligence cannot manage all the maneuvers and services provided by the travel industry. The pilot handles take-off and landing similar to the ways in which travel companies continue providing differentiated experiences for travelers. Throughout the flight and the travel experience, however, intelligence helps to optimize for personalization, scale, and convenience.
