Can you name what technologies fall within the realm of artificial intelligence (AI)? If you aren’t sure, you aren’t alone. According to our research report, “Outlook on Artificial Intelligence in the Enterprise”, 88% of respondents claiming they do not use AI technologies went on to cite using specific solutions that rely on AI techniques. Talk about a definitional problem.
At Narrative Science, we view the AI ecosystem from a data-driven perspective. We think this perspective is important because it ultimately determines which technology is a best-fit for a business problem even if the end-results are similar.
In the IIA’s "Introduction to the Intelligent Systems Ecosystem", Chief Scientist Kris Hammond, states it well when he says:
One of the more difficult issues associated with these emerging intelligent systems is that while each is built on a very different technological foundation, they tend to be bundled together and are often viewed as indistinguishable. No one cares if one system organizes data as a tree and another uses a table if they both get to the same answer at the same time and scale in the same way.
Unfortunately, even though they are often bundled together, these emerging intelligent systems not only work in different ways under the hood but also present us with very different features (and requirements) once we sit down in the driver’s seat.
8 Categories of the AI Ecosystem
So from this perspective, the AI ecosystem can be divided into eight general categories -- although there is some overlap based on the applied algorithms. Those eight categories are as follows:
Evidence Based Reasoning
Machine Learning Systems
And of course -
NLG Possesses the 3 Key Functions of AI Technology
Breaking it down further, the eight categories of technologies deliver one, two or all three of the main functions below:
Situation assessment: Building a characterization of the current state of the world.
Prediction: Extending the current state into a set of predictions about what will happen next.
Advice: Given a set of predictions, providing advice as to how to respond based on a set of goals.
Natural language generation is unique in that it is the only system of these that makes "use of all three components in the core dynamic to first assess situations, draw conclusions from those assessments (including predictions) and then transform those conclusions and any advice that is linked to them as language. Agnostic to the source of the data it uses or derives, this is one of the few technologies in the ecosystem that makes use of all three of the 'assess, predict, and advise' stack," (as summarized by Kris Hammond in the IIA research brief mentioned above).
Humanizing Artificial Intelligence
Another way of looking at it is that Natural Language Generation is the communication layer between man and machine. The ability to audit reasoning within an intelligent system via natural language will be vital to the adoption of these systems as all users, not just the most technologically-skilled, will need to understand the why and how behind the technologies’ decision.
In that way, NLG helps ‘humanize’ artificial intelligence, increasing transparency into how these systems operate and thus playing an essential role in helping enterprises trust and rely on AI systems. NLG isn’t just another AI technology -- it is the literal ‘voice of the machine.’