The poster child of Artificial Intelligence (AI) is the self-driving car. Uber, Google, and others are pouring billions to get them to work, but this doesn’t mean AI is beyond the reach of ordinary businesses. Many of the building blocks are already in place to implement smaller-scale AI-driven solutions without requiring either colossal investment or an army of data scientists.
When we ask an Alexa a question, for example, there are two things going on. The first is converting to speech to text, and the second is deriving your intention. Speech conversion is solved by AI algorithms, but determining intention is more complex and the topic you need to customize for your business. The biggest challenge here is thinking of every different way a user can phrase the same intent (for a cinema: When’s Parasite on? Are you showing Parasite?). Once the intention is determined, making an Alexa skill is actually very simple.
Another area to consider is text prediction. User convenience tools such as LinkedIn’s messaging system suggests quick responses to inbound messages using AI algorithms. This has learned from thousands of message interactions and mimicking similar conversations. Many businesses that offer a means to search for products embed this technology in their experiences, enhancing their search with an autocomplete selection that anticipates the next words a user may type. This kind of anticipation makes a better experience for your customers.
The reality is that most AI, presently, is better harnessed enhancing routine tasks and sorting through massive sets of data, but there are some innovations in the area of product optimization that suggest a bigger step change on the horizon. A UK Restaurant chain, for example, has taken anonymous point-of-sale information and blended it with details such as the time of day and weather to predict trade and how menus can be adapted to increase profit. When bookings are anticipated to be low, staff are advised to promote dishes that take longer to consume, and which customers accompany with high margin drinks. Machine Learning has provided businesses with the power to analyze thousands of data points to determine what factors are drivers of the behavior that they are seeking to optimize.
It’s not often clear which AI technologies can best optimize a business, but asking whether an expert with sufficient data and time could solve the problem is a good place to start. If the answer is that the expert could not solve the problem, then there is no magic bullet. One also needs to be aware of the limitations of machine learning; there is still bias in areas such as facial recognition, as by its very nature that technology which learns from data you give it. If you do invest in such innovation, you need to carefully assemble representative data for it to work well, not just the easiest data to get.
We are some time away from AI that can truly think like a human — but changes have incrementally been impacting behavior in recent years. As these changes are introduced, those implementing them will gain a competitive edge, so it is important for businesses to stay ahead of this curve, in both technology and training, to avoid missing out on the possibilities delivered by AI.