US Army Will Develop AI That Is Able to Predict World Events
The Darpa agency announced on January 4th that it is working on an AI that is capable of predicting world events for national security purposes.
Join the DZone community and get the full member experience.Join For Free
It is within the Kairos program (Artificial Intelligence Reasoning Over Schemas) that Darpa is working on this Artificial Intelligence that it believes can predict world events.
The model chosen to implement Kairos is the use of schemas. This concept, defined by Jean Piaget in 1923, consists of offering contextual and temporal reasoning on complex events. Basically, it is a matter of combining a series of events to produce a final event. A diagram thus describes the organization of an action corresponding to the structuring of an action. That is how we can say that someone who leaves home on the first Saturday of the month — with his wallet, a bag, and who is heading for a shopping mall — will most certainly commit... his races of the month! Okay, the Darpa is not very interested in the example, but I'll let you see the logical path for a terrorist.
So let's take the case of a terrorist, then. We know that terrorists often first have a period of "agitation" on social networks, meaning a strong appetite for articles that are critical of the government, the system, religion, or whatever bothers them. Without following a long period of calm, then shortly before their attack, the terrorist partakes in some unusual events (high spending of money, new trips, will again have small activity on social networks). In short, yes, this modeling has an interest in detecting terrorists who will take action, even if it will be necessary to see the right percentage of detection over time.
Nevertheless, I doubt that a simple schematic analysis can predict everything. For example, it is difficult to predict popular revolts. In my country, France, I have been hearing since I was born that there is going to be a revolution. I am still waiting even if the events that are underway in my country may amount to a popular revolt. And the trigger for this revolt, which mobilizes several hundred thousand people, was an online petition calling for a reduction in fuel taxes.
Moreover, it does not seem to be a long time analysis tool but a short time analysis tool. Predicting the economic collapse of a country like Greece could not have been predicted a year ago; no indicators were found.
Opinions expressed by DZone contributors are their own.