AI is capturing attention as a transformative technology for enterprises. Fundamental to AI is the use of ontologies, says Seth Earley, CEO of Earley Information Science (EIS), a consulting firm focused on organizing information for business impact. His new book, The AI Powered Enterprise: Harness the Power of Ontologies to Make Your Business Smarter, Faster and More Profitable, due out in April, focuses on the importance of ontologies as a foundation for AI success.
BDQ: How accessible is AI becoming?
Seth Earley: Organizations do not need to hire expensive data scientists to get value from AI. The large technology vendors as well as many startups are offering “AI platforms” such as IBM’s Watson Studio, Google AI, and Azure from Microsoft. These development environments have configurable solutions (like simple chatbots) as well as the ability get to the more sophisticated tools and frameworks. What this means is that AI will be democratized with greater access to sophisticated algorithms for organizations without deep in-house data science expertise. This will also create “data oligarchies” with a great deal of power in the hands of a few companies, because in exchange for using their platform, they have access to more and more data.
BDQ: Why is it important for organizations to start to prepare for AI now?
SE: There are a couple of classes of applications that we can consider in terms of preparing for AI. The first is doing things in a more efficient way and helping to make sense of, and get insights from, large amounts of data. The second class of applications includes new ways of doing things—for example, streamlining the user experience through personalization or responding to real-time field performance of products.
But, regardless of the class of applications—and the delineation is not hard and fast because certainly there are overlaps between them—organizations still need to do the basics. If product names or categories are missing or inconsistent from application to application, that data will need a lot of cleansing and manipulation to be usable for analytics or machine learning and AI. With AI, quality data is the price of admission and therefore the unsexy things like governance processes, quality assurance processes, and data ownership and data curation have to be in place to drive these programs.
BDQ: How will AI change the way companies operate in the future?
SE: AI is an extension of present-day technology that is used to reduce the cognitive load on the user or improve how we interact with technology. AI can also solve problems that were not readily solvable by traditional programming approaches. AI algorithms can process very large amounts of information, find patterns, and make predictions that were not easily addressed with traditional computing technology.
BDQ: What is the biggest inhibitor to ROI in AI projects?
SE: There are several sources of problems with ROI in AI projects. The first is that people don’t clearly identify what they’re trying to accomplish. Many organizations are experimenting, trying to figure things out. AI success and ROI first depends on the clarity of the problem that you’re trying to solve, rather than trying to deploy shiny new technology.
Another challenge is that it’s very difficult to go from a proof of concept or pilot to something that’s operational. When you do a pilot to show something working, you are hand-curating the data, fixing it, and putting it into an algorithm. The results in the pilot may be great, but then when you try to deploy it, you don’t have the same luxury of crafting in the data processes where you’re treating the data with loving care by a data scientist and have a lot of data janitors cleaning up the data.
Other inhibitors include problems with hype and unrealistic expectations of the tools. There may also be a misalignment of the tool complexity and the type of problem. For example, IBM’s Watson technology is very powerful. But there are certain applications that you really don’t need it for—and, if you try to apply it, you’re using a sledgehammer to kill a flea.
BDQ: Why should organizations be focused on ontologies?
SE: Ontologies are the knowledge scaffolding of an organization. They are the structure and the framework on which the organization processes its data content and its knowledge insights. The ontology really becomes the soul of the business, because it contains relationships of condensed clear definitions of the business concepts and how those business concepts are implemented in multiple contexts and applications.