When IBM’s Watson supercomputer won Jeopardy in 2011, cognitive computing exploded onto the public scene as a very possible — if not slightly unnerving — reality. The industry has been interested in the prospect of computers that think since science fiction comics of the 1950s first tantalized readers; Watson, however, let the general public see firsthand just what was possible.
Today, technologists are looking past simply winning trivia game shows toward a future where machines are more than just computational — a future where they are able to put together seemingly unrelated data to make decisions (under some level of human supervision, of course).
While the images of Stanley Kubrick’s 2001: A Space Odyssey, in which an intelligent computer famously thwarts a human’s attempts to power it down, spark an unavoidable conversation, those working in the cognitive space say there is real potential for governments in a range of applications.
During the Florida Digital Government Summit* Friday, May 13, IBM's Tim Paydos discussed the use cases and what the “thinking” machines' path forward looks like.
Among the many applications, Paydos, the company's vice president of Worldwide Cognitive Solutions, said the tool provides a real value in the criminal investigation space. Following the Boston Marathon Bombing in 2013, and 2015 and 2016 terror events in Europe, Paydos said the cognitive technology proved to be a valuable resource for investigators.
“When you think about the Boston Marathon Bombing, it was very much an analytic challenge and an opportunity,” he said. “Within 24 hours of the detonation of the event, we had acquired 480,000 discreet images and videos of the site, just prior to the detonation, just after and coinciding — with petabytes of information. The challenge is in ingesting all of that information, and then having someone or something go through all of that imagery to find the bad guys. The opportunity, however, was in that body of data, that corpus of data, lay the information we needed to go after the bad guys.”
The ability to match seemingly unrelated information and aliases to specific individuals has also been of value to the law enforcement and those in the national security space.
According to Paydos, the ability to move past querying systems based on a set of pre-determined questions opens the door to new connections that otherwise might have gone unnoticed by analysts.
What’s more, he said, is that analysts often clean the “dirtiness” out of the data in an effort to standardize it for analysis, which can strip out the useful bits of information.
When put into different contexts, cognitive technologies have also helped to answer daily operational problems at the municipal level. In the city of South Bend, Ind., more than two dozen overflow events at the sewage treatment facility earned substantial fines when the wastewater reached the adjacent river.
Through the application of cognitive computing and systems throughout the plant, officials determined more efficient routes for untreated wastewater and avoided the costs associated with a $120 million overhaul project.
Despite the benefits of smarter tools and tactics, Paydos said a deliberate approach is required when entering the cognitive space.
“The irony is that while cognitive analytics will be transformational and revolutionary, the path to success is iterative and evolutionary,” he said. “Do not go out to build the big, intergalactic cognitive analytic architecture and stick it in the cloud.”
Instead, he said, focus on current capabilities and meeting the high-priority business goals of your organization before investing in shiny new tech.
*IBM was an anchor sponsor of the two-day event, which was hosted by Government Technology and its parent company e.Republic.