With AI/ML, context matters, especially in the Airline Industry
While embarking on artificial intelligence in the airline business, most of us assume that you need vast amount of datasets to get started. No doubt that AI is fueled by data, and so it only makes sense that the more data you have, the smarter your AI gets, right? Not exactly.
How does one carve out actionable intelligence by applying AI to data? Here is where, context matters in our industry. To put it in layman’s words, one can build the biggest dataset imaginable, but if you don’t know what to look for and you don’t have the ’right’ data to do it, then you will land up in a no man’s land.
AI is not some wand where one says Abracadabra and you get cues that can be used to execute a set of actions that can move the needle resulting in measurable value proposition. AI is a combination of algorithms, technologies and processes, each with a specific, fine-tuned purpose. Airlines first need to zeroin on the impact they want to see. This will help them to determine and focus on curating the datasets to those objectives that have the best opportunity for success from AI.
Let us take the example from the air cargo industry. It’s interesting to know how the United States Postal Service (USPS) automates mail sorting. With the help of machines and advanced optical character recognition (OCR) technology, the USPS can now read and process 98% of all hand-addressed mail and 99.5% of machine-printed mail without human assistance. By linking this technology with a relatively small and finite data set of U.S. zip codes and cities, the USPS can now process upwards of 36,000 pieces of mail per hour. With the USPS facing harsh financial challenges in recent years, the impact of this automation is immeasurable. (source: facts.usps.com)
Another interesting example from our industry is the usage of small, high precision data to make enormous gains with AI from Boeing. In 2015, Boeing launched the Aerospace Data Analytics Lab in partnership with Carnegie Mellon University to develop AI technology for airlines. One such project aims to dramatically reduce maintenance costs with AI by standardizing maintenance logs. (source: wired.com, cmu.edu)
As you all know, every aircraft is required to keep highly-detailed maintenance logs. The language used in different parts of the world makes it tough to decipher and process these logs. From there, it only gets worse. Some logs are captured digitally; others are hand-written. Some maintenance workers stay in the lines, others write notes and abbreviations in the margins. Just imagine the plight of an average maintenance worker, translating these variations on the fly which can be next to impossible. But with AI and a narrow data set of common aircraft maintenance terminology, it becomes possible to capture and dynamically translate these logs in real time. By leveraging AI to improve the speed and accuracy of the airline maintenance workflow, airlines stand to save billions.
There are several examples like these in our industry on how AI powered by precise data can lead to a huge impact. How can one put these ideas to make it work at your airline or any company for that matter? Here are some basic guidelines:
Set goals that are tightly tied to the hip with the business objectives. The first guide line is setting goal(s) with the team across disciplines that is tightly integrated to the business objectives. In the case of AI, it is critical. Due to the prescriptive nature of AI; the more you can taper the business objective, and the more contextually precise your data set, the more likely you are to get some meaningful results. There has to be a champion in the organization that has visibility across the cross functional teams. This is essential as only the objectives are taken seriously and heard by those who are not directly part of the endeavor. One would want to put together a team that has stakeholders from various verticals including finance, sales, operations and someone from the executive level as well. This will result in identifying bottlenecks and also opportunities and it will be faster to find a practical solution to solve them
Taming of the Shrew. As Bard himself said, “No profit grows where is no pleasure ta’en:
In brief, sir, study what you most affect.” Almost all airlines have a huge dataset with great value to their business. There is a disconnect between the perceived value and this dataset itself. This data has to be clipped for outliers and cleansed thoroughly for accuracy. As most data analytic practitioners would say, “junk in, junk out”. So this step is something that has to be executed with great finesse to make the data actionable. While calibrating this data, one can create a useful framework for taming data chaos and extracting small high precision data focusing on the lifecycles of customers, spending habits, origin – destinations, seasonal events and 3rd party vendors. Following the lifecycle shows you all of the steps, systems, and stakeholders involved. You will most likely understand your business better and find gaps as you examine this lifecycle. The gaps will also lead you to areas where you are leaking value. These are your opportunities to make a clear and measurable impact. Focus on the key data surrounding these gaps and you will have more precise and actionable data.
Picking the right technology for the goals. There’s a lot of buzz right now about machine learning and AI — and it’s justified buzz. There are quite a few impressive technologies with great promise for any level of executive for any sized airline. They are also now available at a fraction of the cost compared to even a few years ago. Don’t hire a team of a hundred data scientists; look to the growing ecosystem and pick the right tool for the job.
In our world of transportation, airlines (both pax and cargo) are always looking for big bang solutions — some breakthrough that can give them an edge. But the reality is that when you get practical, you can start accumulating lots of low hanging fruits and smaller wins — and you can do it quickly. Over time, this accumulation can drive massive outcomes.
In my opinion, this is the right way to think about AI. It’s not a magical wand— it’s a highly-specialized set of tools. It’s not about shooting for the stars — it’s about winning one battle at a time. And it’s not about amazons of data — it’s about small, high-precision data.