What is Happening?
You would need to be hiding under a rock to not have noticed the recent rise in popularity of the term “machine learning.” This statistical method, which consists of using computers to build complex and sensitive predictive algorithms, underpins much of what today is being billed as artificial intelligence. While machine learning has been around in various forms for many decades, improved access to computational power in the cloud – as well as the application of machine learning capabilities by large internet companies such as Amazon, Google and Facebook to improve recommendations, searches and content filtering – have made it top-of-mind for businesses engaged in Digital Transformation. Depending on how it is applied, enterprises can use machine learning to improve targeting and interacting with customers, to better automate tedious tasks in back-office processes or to help model and avoid financial risk – a very flexible tool indeed.
Figure 1 – Google Trends Data – Machine Learning. Source: Google Trends (Accessed 20 April 2017). Note: Google Trends data is represented as Search Interest in the given term over Time, with 100 indicating peak popularity and 50 indicating the term was half as popular.
As seen in Figure 1, the growth in interest in this topic is clearly visible with the search instances roughly doubling over the last year.
Why is it Happening?
The fact that businesses, software engineers and statisticians are using a new method isn’t particularly noteworthy in and of itself, but, when taken in context, it represents the beginning of yet another important shift in the evolution of computing. Until now, the evolution has happened in three epochs, starting from early tabulating machines and progressing to the extensive mobile and social networks of today. The series of developments goes (roughly) like this:
- Systems of Record – developed primarily to store data about the business
- Systems of Interaction – developed to provide new ways to work with data in the Systems of Record and to focus on creating, reading, updating, deleting – or “CRUD” operations
- Systems of Engagement – developed to move beyond a strict transactional system to add share and react functions on top of CRUD operations. These systems formed the basis of collaboration and enabled a level of abstraction from the transaction to better allow users to understand the process and engage with it.
Each of these developments has been built upon the last, and none of them replaces the ones that came before it. Instead, they offered additional abstraction from the data and the computer’s processes to get closer to what the human processes needed to be. While the nature of these systems is not predicated on the evolution of databases, the concomitant evolution of databases from tabular > hierarchical > relational > graph and unstructured has also enabled many of the additional capabilities of these systems over time
As Systems of Engagement have matured over the last few years, we have been watching closely to see where the innovation was likely to lead next. Based on our recent survey and interview programs, we believe two more systems will be added to the three existing ones, and that they will be:
- Systems of Understanding
- Systems of Intelligence
The surge in interest in machine learning and the increasing maturity of predictive analytics technologies already are building this nascent group of systems to help us understand the world around us. “Cognitive” systems, as Systems of Understanding are frequently called, include many sophisticated tools that help expand the scope of both human and computer decision-making by giving us new ways of understanding data and the relationships therein. The computing systems being developed now build on top of Systems of Record, Systems of Interaction and Systems of Engagement to help us describe, correlate, and predict based on information to which we have access.
While Systems of Understanding are still immature, we are seeing early evidence of a future of Systems of Intelligence, which will be defined by the ability to understand within a real-world context. This capability will allow computers to navigate beyond describing correlation to inferring causation.