The New Marketing: Machine Learning, Deep Learning & Artificial Intelligence

Artificial intelligence and machine learning have been around for a while but their use for facilitating marketing related decisions is only now picking up pace. The forces which made this possible are faster, cheaper and more powerful parallel processing on the one hand and infinite storage and better algorithms on the other.

Approaches have evolved too. Rather than teach machines everything we know, we now code them to learn. In fact, a sub-field of machine learning called deep learning (also referred to as deep structured learning or hierarchical learning) is based on learning data representations as opposed to task-specific algorithms. The learning can be supervised, partially supervised or unsupervised. Deep learning, as a relatively new field, is in fact one of the ways machine learning is getting closer to its aim – artificial intelligence. Indeed, the cutting edge of A.I. is “deep neural networks.” These are mathematical algorithms that learn patterns on their own by analyzing data (referred to as unsupervised learning), in addition to having the capability of being trained by humans by associating input to output (supervised learning).

Access to vast amounts of data on the Internet is fueling this trend…. Indeed, machine learning is the use of algorithms to parse data, learn in the process and then to help make choices or predictions. But deep learning, a technique for implementing machine learning, has been taking this to a new level. The results are profound and far-reaching.

Most of us have been familiar with supervised learning, where the output of an algorithm is already known. However, unsupervised models are taking AI to a new level. Instead of establishing the process necessary to get from your input to your output, through teaching the algorithm from a training data set, we now enable machines to learn iteratively.

These “unsupervised models” do not require a human to identify the ‘questions’ or tag a training set. They simultaneously learn both the questions which best capture the qualitative information and what the answers to those ‘questions’ should be for a given input. These could be a sentence, a video or any form of unstructured data.

At the broad societal level, AI and machine learning will transform the very nature of work. Any tasks and processes, which can be automated, will no longer require human effort. That much we’ve known. But even tasks, which require complex reasoning and intuition, will eventually lend themselves to artificial intelligence, however hard that may be to grasp. Many predicted that IBM’s Watson would never beat a chess grand master but we now know otherwise. In 2016, another bastion of AI denial (the world of Go players) saw that defeat of Korean Go Master Lee Se-dol to Google’s Deep Mind AlphaGo program. Actually, Go exemplifies "reinforcement learning" that aims to maximize the expected cummulative reward for a given task (win a game or get the best possible score). It has to do with simulated sequential decision making (such as planning) towards a goal (such as maximum efficiency, profit) with huge implications in business contexts.

In marketing, there are several areas where developments are transforming the very ways we reach and affect customers. Take for instance, natural language processing for sentiment analysis - sentiment classifying algorithms based in natural language processing can crunch millions of social media posts.

Another frontier is image processing. Social media monitoring tools based on text analytics and NLP only addressed part of the picture (no pun intended). It has since become clear that for more robust analysis and insight, we need to deal with the vast array of images which are on the web. This class of capabilities is now being deployed to address: logo detection, text extraction, object recognition and the more difficult skill to be able to caption an image with a full sentence that describes its theme. The emotions in an image can be detected by using facial recognition technology.

Michalis Michael, CEO of Digital MR noted in a recent blog post (https://www.digital-mr.com/blog):

“Up to now, apart from having the ability to access the relevant posts, text analytics and Natural Language Processing (NLP) have been the main disciplines required by social media monitoring tools. That is not the case anymore. Tweets with an image get retweeted 150% more than those without one; they also get liked 89% more. According to Twitter, 77% of all tweets about soft drinks do not have a textual reference to a soft drink brand or anything related to the product category.”

We know that artificial intelligence will play a key role in all aspects of our lives, well beyond marketing. What we don’t know is how benign the effects will be and the kinds of challenges this exciting yet scary new capability will entail.

For those in marketing, though, the exciting new capabilities pose huge opportunities and a chance to gain competitive advantage!