Google, Bing and Technological Singularities
The way that humans manage and process information has fascinated me for a long time. There are a multitude of interesting and challenging issues and, like a child in a candy store, every savory morsel attracts my attention. So it isn’t surprising that a comment about “technology singularities” by futurist Ray Kurzweil in Time magazine triggered my thoughts on how developments in artificial intelligence relate to information management and, in particular, to my favorite subject asset information management (AIM).
AIM is challenging for a lot of reasons, but the extensive interrelationships that exist between different pieces of information is one of the most interesting. Modeling and managing these relationships is critical for enabling users to efficiently access information, navigate to related information, and maintain information quality. And, the issues that we face in AIM are similar to what AI researchers face in understanding how humans manage and use information.
Sophisticated information management is natural for humans. We create our own models of how the world operates and use new information to refine and adjust these schemas. In the process we assign meaning to new information that enables subsequent recall and processing from a variety of perspectives. In most cases, this recall is quick and efficient in terms of the information we retrieve. In other cases, we only get it to “the tip of our tongues” and either give up or struggle to find some other cue to unleash what we know. In such cases, we know that the information is there but we just don’t have the relationships to get us there from the original cue. (An interesting aside: I recently learned that one benefit of getting old is that experience enhances our information modeling skills and indeed makes us wiser. But, our ability to retain information declines over age, so this benefit does not always reveal itself. )
We can compare this with the state of information technology today by just looking at some of the products we all use. Google is obviously a powerful tool for finding information. It allows us to locate millions of items that relate to our search terms. But, as the excellent Bing ads have taught us, Google recalls everything that is even remotely related to our query. Ordering by relevance, based on things like popularity and number of common terms, certainly directs us to the most popular information, but the simplicity of this data model does little to really relieve “irrelevant information overload”.
While I don’t use Bing myself, Qi Lu, the president of Microsoft Online Servcies (and an alumni of my own alma mater Carnegie Mellon University), tells us that it will help us by finding only the relevant information. They do this through natural language parsing of our queries to extract the “semantics” of what we really want to find. This can certainly be helpful, but it still builds the data model after the fact, so information retrieval by keywords remains and this offers little help in finding other related information.
Kurzweil, whose comment motivated this discussion, apparently believes that advances in miniaturization and computing power will enable us to have smart chips embedded in our brains that will constantly monitor our thoughts and retrieve any information we might need before we conciously know that we need it. In essence, this will tie our internal data models to the cloud, so people who think they know everything really will. Obviously, this will also be a godsend for the elderly, who will finally be able to display their wisdom.
Obviously, the challenges we face in AIM pale in comparison to those I just discussed. Not because they are different, but because AIM is much more constrained. So it should not be surprising that we have the knowledge today to build rich data models for asset management and use this to ensure information quality and effective access for every stakeholder. And, organizations that have begun this journey are seeing significant benefits like: role-based information access that can truly filter irrelevant information and protect information from unauthorized eyes; and, fast information navigation through 3D models of facilities that marry critical spatial relationships with the user’s knowledge to enable intuitive information access and provide valuable context for retrieved information. We might also say that smart maintenance technologies have already enabled an AIM singularity as sensors and software tell us what we need to address, trigger automatic recall of relevant information, and even initiate actions to address the problems.
While I am not too keen on the idea of brain implants, I would really like to be around for the world that Kurzweil envisions. But I appreciate the enormity of the challenge, and temper my hopes that anything like this will ever occur in my lifetime. On the other hand, I do expect to see AIM advancements unleash incredible benefits in the asset management world, and I am thankful to have had the chance to participate in these developments.