In the previous installment, we discussed some of the basics of Artificial Intelligence (AI) and explored some relatively simple examples of how to weave the technology into key business processes by analyzing and predicting user behavior. In this article, we’re going to dig a bit deeper into AI-implementation. We will take our airline use case a step further, and we will describe a specific example of how EastBanc Technologies solved a particularly challenging problem through AI and machine learning.
First, let’s look at where to start your AI project. Many businesses already have access to the data needed to begin implementing AI - even if they’re not yet taking full advantage of it. So how to put that treasure trove of data to use? Every customer or user leaves a data trail; all you need to do is follow the breadcrumbs. Then analyze and learn from them to identify key trends and, by association, business opportunities. Once a potentially interesting breadcrumb trail has been identified, you’ll need a data scientist who understands computer science, machine learning, and deep learning to make sense of the data.
The data scientist will apply a simple algorithm on top of it to identify patterns. From that point, AI will continue to open the door to new opportunities. To illustrate this, let’s return to our airline example. We’ve already covered how tracking, analyzing, and predicting passenger behavior can help the airlines maximize secondary sources of income, such as duty-free shopping by serving intelligent, targeted offers based on behavioral analysis. So let’s say the airline in our example delves deeper into the buying patterns of its passengers. The algorithm detects that one of its most frequent fliers buys a box of luxury Swiss chocolate every third return flight from Europe – and always when flying from Switzerland. Similarly, that same passenger buys six bottles of Abbey Ale every second return flight from Belgium. Maybe when flying from Central America or the Caribbean the frequent traveler buys a bottle of Cuban rum every third flight. Evidently, this traveler particularly enjoys high-end Scotch Whisky: A bottle of 24-year-old Single Malt is bought duty-free on nearly every international flight. Getting these insights is easy and fast. Every pattern can be tracked and analyzed.
In a more advanced example of AI-implementation, an EastBanc Technologies client asked if AI could perhaps predict the probability of winning or losing certain legal cases based on historic trends and patterns? Too good to be true? Not at all.
Using Minimal Variable Prediction (MVP), we started by focusing on a single initial data point – one particular type of litigation - to build a predictive analytics engine to test the assumption. The engine applied machine learning to automate analytical modeling and find hidden patterns and insights within the data. Next, deep learning was applied. AI was key to extracting specific information from transactional data, such as case wins, losses, and correlating factors, then reporting that information contextually to predict outcomes in future patent appeal cases. The prototype we built enabled the client to refine and iterate to add new data sets and test other inquiries for an ongoing stream of actionable, affordable legal predictions.
Even when you are ready to pursue more complex problems through AI, chances are you won’t have to develop the algorithm in-house (that’s the most expensive part, by the way). We are increasingly seeing sophisticated packaged AI services hitting the market. These are algorithms that solve a very specific problem, such as image recognition. Your team can purchase these and start crunching numbers right away.
As exemplified above, Artificial Intelligence continues to advance the way businesses sell their products and services. Utilizing data to understand a consumer’s purchasing patterns helps AI to know what items to show the person in the future. In return, this will create more growth for businesses via increased sales.
AI, machine learning, deep learning – these technologies are already being used to tackle day-to-day tasks across numerous fields. And when it comes to the true potential of these advancements, we are only just starting to scratch the surface of what might be possible.
Implementing AI and embracing its nigh-endless possibilities is exciting. For many businesses it may also be intimidating But it does not have to be prohibitively difficult or costly. With the right approach and, if applicable, the right partner(s), businesses can start their AI-journey small, and then apply a patient, iterative approach to slowly progress to being AI experts.