About a year ago the CIA created a Directorate of Digital Innovation, its first major new directorate since 1963. Officials called it a milestone in the agency’s history.
Andrew Hallman, who helps lead the unit, said CIA Director John Brennan backed the move. According to Hallman, the agency has improved its forward-looking analysis since it was founded, using complex algorithms and structured reasoning. Deep learning and other machine-learning methods help analysts see how seemingly unrelated data may connect to near-term events.
“This summer was very productive,” Hallman suggested in remarks about the unit’s pace. Analysts can tap CIA data stores, but they also mine public sources. Brennan described the summer as especially busy for the agency’s digital push.
None of this means perfect prediction. It does mean intelligence work increasingly looks like large-scale data science: more sensors, more open web, and models that surface weak signals for humans to judge.
