Artificial intelligence is a tremendously useful tool that can increase productivity and accuracy with certain kinds of tasks. AI has a lot of specific applications that make it very useful for marketing teams and data scientists alike—but it’s important for marketers to understand the basis of these technologies so they can best figure out how and where to implement them into their marketing stacks.
How can AI help me?
As programs get smarter with time and more advanced designs, they can do more things. AI is no longer in its infancy. There are a ton of areas where AI implementation can improve accuracy and speed of programs, and lift the burden of deriving insights off of the marketing and data science teams.
One such area is classification. Humans can do this pretty intuitively. When you walk past a dog on the street, sometimes it can be hard to figure out what kind of dog it is. Other times, there are prominent features to help guide you. Suffice it to say, it’s not hard to tell a Saint Bernard apart from a golden retriever. We can use this same sort of logic to teach computers to observe certain features in a set of ungrouped objects, and start to unshuffle them into categories with other similar objects. Classification is useful for a ton of things in the business world. At Zylotech, we group individuals into segments and microsegments based on particular features we observe in demographics, purchasing data, etc. This is a very complex implementation of the basic idea of classification.
Another area where machine learning shines is predictions. By feeding in features and results from past experiments or observations, you can train a machine to tell you what is likely to happen under various circumstances. For example, you can train an algorithm by feeding it purchasing behaviors to look for cause-effect patterns. Maybe around the time of a big sale, a group of consumers spent a lot more money, or bought more frequently. You can use this information to predict who will buy at the next sale, or vice versa, to recommend a cause that will drive the effect of buying.
Sounds great. I’ll put machine learning everywhere!
Not quite. There are some cases where it doesn’t really make sense to use machine learning—at least, for now. There’s the finicky science of natural language processing, which, while improving, is a little unreliable and subject to the biases of the algorithm designer. Machines aren’t great at common sense, either. If you’re trying to train an algorithm to give you insight into something that lacks concrete data, or that isn’t as straightforward as an input-output relationship, you might be out of luck. A great example of this is sentiment analysis. Modern sentiment analysis gives you a “good enough” result by showing a program what kinds of words and sentence structures connote a particular attitude (for example, “bad” isn’t happy, but “not bad” is). But actual feelings are much more fluid and complex than machines are currently capable of understanding. If your goal is to get a robot to truly understand human emotions, you’re going to be waiting awhile for that kind of breakthrough.
Machine learning and AI are incredibly useful tools to incorporate into your marketing stack. Marketers should be aware of the growing landscape of AI, and work with their data science teams to figure out how and where AI works in their own individual system architectures.
Josh Fayer manages Marketing Communications at Zylotech, where he brings an extensive background in public relations and computer science. Josh Loves music! He sings with an a cappella group, Orange Appeal, (whose recent album Unpeeled he shamelessly plugs at work), and at home he loves to mess around with a variety of instruments. Josh enjoys spending quality time with his adorable dog, and exploring around his new home near the South Shore of Boston.
If you liked this post, check out our other blog post addressing some common frustrations in switching to an AI-based platform.