How Cognitive Computing Systems Differ from Traditional Information Systems

[Transcript of video interview with Sue Feldman, Founder & CEO, Synthexis, at the KMWorld 2015 conference on knowledge, content, and document management.]

Q. How do cognitive computing systems differ from traditional information systems?

A. Cognitive computing systems are learning systems. That means that as you interact with them, they get better. Specifically, if you think about a traditional information system, it's linear. Questions go in, they get matched to documents, and out they come. If you ask the same question two months from now or two years from now, if the same documents still exist in the system, you're going to get the same answer.

In that sense, such a system is predictable, but it also means that it doesn't move along with you in terms of what you're learning, so it's dumb. What you really want is a system to be a partner with you that will move along with your exploration.

A cognitive system handles questions differently. In other words, it enlarges rather than forces you to overly focus a question or a problem statement. This allows you to bump into information that you may not already have thought about, or even asked for. If you're thinking of innovation in particular, what you want to do is to uncover things that you didn't know about. That's been a huge failing in information systems today, because if you're asking a question, it's because you don't know the answer. How are you going to ask about something that you don't know the answer to? You need help, and we have fallen down in designing systems that help people to ask the right question.

The third thing that ís very different is that, because we've enlarged the exploration field--in other words, expanded the questions--we've ended up with a very large data set at the end. That gets filtered for the individual who is asking the question. Who you are, where you are, what you've asked before, what your role is, where you are in the task, what step you're at. All of these questions help to focus that data set down to stuff that's useful for you. And usefulness is what we're looking for, right?

The variety of analytical tools and visualization tools allow you to explore that data set in, perhaps, some unique ways to interact with it. It's a learning system, so every interaction that you've had with the system allows you to improve the system. All those interactions go back into the learning system in order to enrich the information, to add to models that you may have built already. To get the system very much like a person would to learn what it is that you're looking for to help the next person.

I believe very strongly that we're just at the edge of a huge leap into a different way of using information systems so that they become much more of a partner instead of being separated from us by a wall between machine and human.