The future of technology is fascinating. Today we’ll look at the ways emerging technologies will impact technologies we’re already well familiar with. With machine learning getting better every day, there’s a serious need to carefully look into human vs. computer decision-making and what impact cognitive computing can have on the world of document management systems (DMSs).

What is Machine Learning?

Before we get started, let’s consider what machine learning really is. One of the best, most easily understood examples of machine learning is Apple’s Siri. This program looks for patterns in data and can then produce outcomes according to forecast wishes of the user. Is it perfect? Certainly not—there are oodles of webpages dedicated to hilariously wrong advice from Siri. However, Siri is useful and relevant at levels that would have been shocking to comprehend just a few years ago, because she is part of a suite of emerging technologies that facilitate the personalization of computing.

But how does machine learning work? Part of the process involves the development of algorithms that can make changes to a system, and to process how the change in the system changes the behavior of the person operating the machine.  This technology has a million and one possible applications.

How Will These Technology Features Affect DMS?

At the moment, DMS already offers personalization to administrators. Administrators can create and develop workflows along with other file structures within the system. However, this still leaves the question: how will ECM develop personalization on both the departmental level and the individual level, if it’s generally an organization-wide tool?

Personalization With ECM

ECM solutions can help organizations improve one-on-one customer conversations. They can create conversations for increased lifetime value, better conversion, and higher retention over the lifecycle of a customer relationship. Personalization takes this all a step further, in a number of ways:

  • It enhances one-on-one conversations with each and every customer by allowing administrators to better understand their unique needs and past behaviors.
  • It gives deeper insight into every customer’s lifetime interactions and interest. This can lead to the ability to create very personalized content, which may include recommendations for products and even marketing offers in a preferred communication channel.
  • Personalization allows organizations to reduce communication to only offers that are made to customers who may be interested in their offer. It can help create intuitive, marketer-driven controls and rules.
  • Organizations can better understand the impact of various programs on conversions by taking advantage of dashboard reporting.

The Limitations of Computers that Can’t Think

While computers can learn a lot, the reality is that human vs. computer decision-making is very different. Cognitive computing obviously does not literally mean that computers can think—they can’t. What computers can do is learn. They can learn very specific rules and how to respond to a wide range of decisions, but it’s still up to administrators to create those roles and designate those decisions.

For example, a recent article in the Wall Street Journal described a computer that worked scheduling meetings. It could schedule meetings between various parties, respond to emails, and interpret text responses from individuals. In the end it was able to find a time that worked for everyone. The reporter who worked with the program described at as if they were working with an assistant.

The program could respond, it could ask for different times, and it could clarify the time and date. However, this is a very narrow use of machine learning. It is being applied in many ways, for example, chat bots are used in call centers and customer service centers to move calls to the right party, but computers can only emulate human thinking when the domain is narrow.

How to Get the Best Results From Computer Decision-Making

The best results from cognitive computing come in situations like the one described above: when specific rules are present, when the domain is narrow, when the tasks are both well-defined and procedural, and when both the user’s context and situation are easy to infer based on who they are and what they’re doing.

In many cases the most difficult decision to make is whether taking time today will save time in the future. If a company is handling a situation that comes up frequently, then it may make sense to take the time to teach a computer to handle it. On the other hand, less infrequent situations can be much easier to handle by a human than to train a computer to handle. The gray areas in between are the areas companies have the most trouble with.