Introduction
The prospect that artificial intelligence might transform intelligence work is not new. In 1964, CIA was worrying about Soviet advances into “artificial-intelligence” and their national security implications. This early AI research was similar in nature (though not, of course, sophistication) to contemporary goals and techniques, including pattern recognition and machine learning. CIA observed, for instance, that the Soviets were using supervised machine learning to rapidly assess the seriousness of burn injuries. As AI technology has matured and diffused throughout our society, consideration of how to improve intelligence with it has quickened. This work has focused on the ramifications of AI for intelligence analysis.
Much less attention has been paid to the ramifications for human intelligence (HUMINT) operations. But frontier (i.e., state-of-the-art) AI is already changing how HUMINT collectors—case
officers (COs)—do their jobs. And some people believe that emerging technology, including AI, threatens the entire HUMINT enterprise.