- By Jenna McLaughlinJenna McLaughlin is an intelligence reporter for Foreign Policy, focusing on the culture, dynamics, and events happening in the National Security Agency, the Central Intelligence Agency, and the other 15 members of the intelligence community—plus the way the sensitive information they gather and analyze informs and directs the White House and policy makers on the Hill. Previously, McLaughlin was a national security reporter for the Intercept where she covered everything from the FBI’s secretive subpoena powers to cybersecurity companies in the Middle East. Before that, she covered similar topics including the rise of the Islamic State at Mother Jones Magazine. You can reach her with tips and responses securely through Signal or WhatsApp at 203-537-3949, or through her email, email@example.com.
The CIA currently has 137 different artificial intelligence pilot projects underway, according to a senior agency official.
Dawn Meyerriecks, the CIA’s deputy director for science and technology, told an audience at the annual Intelligence and National Security Summit in Washington that the agency has a “punch list” of different artificial intelligence problems that it wants the private sector to work on. The CIA is already coordinating this work with In-Q-Tel, the agency’s venture capital firm, she said.
The intelligence community has been eyeing artificial intelligence and machine learning to replace some of the tedious tasks its analysts perform for a while now. In June, Robert Cardillo, the director of the National Geospatial-Intelligence Agency, vowed to bring in robots to do 75 percent of the tasks currently being done by employees to analyze and interpret images beamed in from feeds around the globe and in space.
The CIA and other intelligence agencies have been investing in artificial intelligence for decades, and the technology is already prevalent in certain security applications including facial and voice recognition. But there are a growing number of more ambitious practical applications, including in detecting malicious hacking online and helping pilot drones and other autonomous vehicles.
The transition from analyst to robot won’t necessarily be seamless, however. Replacing analysts with algorithms is troubling to some who believe humans are irreplaceable in the delicate art of intelligence. “We can’t just feed [information] into a black box,” Meyerriecks said. “We can’t go to leadership with recommendations when we don’t understand what happened in the middle.”
But intelligence officials are warning that adversaries and competitors like Russia and China are competing for similar capabilities.
Some technology that was proprietary even 10 years ago in the United States is “mass-produced in China today,” Brian Sadler, a senior research scientist for intelligence systems for the U.S. Army, said during another panel at the conference.
Meyerriecks, however, said she was confident in the United States’ long-term ability to outpace its opponents. “If there’s a bear in the woods, I just have to be faster than the slowest person,” she said.
Photo credit: CHRIS McGRATH/Getty Images