Technicalities aside, there is one crucial thing you will need to bear in mind when it comes to Artificial Intelligence (AI): It needs to “learn” before it can do something—anything—for your organisation. This is true for both narrow AI, which can perform one specific task very well, and, most especially, general AI, which can do a variety of tasks such as language comprehension, problem-solving and object recognition.
To understand this dynamic better, consider actual human intelligence, which needs to learn how to do a task first before it can accomplish it correctly and repeatedly. And for learning to take place, there has to be some form of teaching, in which the specifics of the task are explained in detail—from background information about it to how the task is supposed to be done.
The same process enables AI to be “intelligent,” and it is made possible by machine-learning and Deep Learning. The former takes in various data-rich algorithms to learn how to process information based on all available data; the latter, on the other hand, builds on machine-learning by cascading what it has learned into different layers to enable abstraction, as when AI, for example, adds information to a given category.
Context Matters—A Lot!
In keeping with the human intellect analogy, it should be pointed out that context—prior knowledge and experiences, dynamic transactions with other humans and the environment and internalisation of cultures and societal norms—is central to human learning. The reason being is that context determines how people discern the important from the unimportant, in turn leading them to make decisions using information relevant to a given situation.
The same is true for AI, which also needs context in order to be humanlike in terms of making decisions—that is, based on information, depending on the situation and for the best outcomes possible, or at least the right ones. In fact, AI needs far more context to learn so it can better approximate the decision-making process of humans.
Knowledge Is Context, Knowledge Is Power
This need for context is where knowledge graphs come in. A knowledge graph is a specific type of graph that emphasises contextual understanding by presenting interconnected sets of facts that describe real-world entities, events or things and their interrelations in a format understandable to not only humans but also machines (or AI). It uses an organising principle that provides meta-data, which then adds connected context to enable both knowledge discovery and logical, information-based reasoning.
Knowledge graphs, therefore, are a way for AI to learn because it presents a wide array of information revolving around relevant attributes and with connections between and among data sets already apparent. In other words, information is presented by knowledge graphs in such a way that AI can just “look” at it and then learn from it—much like someone looking at a graph and deciphering the message it is trying to impart.
Knowledge graphs are thus considered “context for decisions.” They can be context-rich, where internal knowledge documents and files are graphed and meta-tagged; external-sensing, where external data are aggregated and mapped to entities of interest; or natural language processing, where technical terms, acronyms, abbreviations, misspellings and more are graphed.
Each type of knowledge graph has specific use-cases and the use of one or the other will primarily depend on why an organisation is deploying AI in the first place. All three can even be used in harmony with one another in case AI is being used for a variety of purposes across the organisation. That would, in fact, be ideal as they will be able to provide the needed context and insights that can enrich AI—and help unleash it fully.
Build AI With Context Using Graph Technology
If you want to maximise AI, you must give it contextual power using graph technology. And you can do that using the Neo4j Graph Platform, which will enable you to build intelligent applications that will help you address challenges, find new markets, enhance the customer experience, improve business outcomes and, yes, develop an enriched, fully functioning and intelligent AI.