The Defense Department has become fascinated with Artificial Intelligence (AI), and rightly so, as this technology will be as transformative as electricity was a century ago. This paper proposes that the convergence of AI and information operations (IO) represents a greater strategic liability than computationally similar applications—physical adaptations, like drone swarms, and informational applications, such as intelligence process optimization—that have drawn so much budgetary attention. Using a hypothetical “Influence Machine,” the article explores evolving techniques that achieved near-human acuity on many influence tasks. With all the necessary data essentially pre-structured, machine learning can perform these tasks at a massive scale. Using the historical parallel of the impact of the printing press during the Reformation, the author unpacks the exponential potential of emerging influence systems. He also examines how, applied during a time of crisis, such tools may provide a strategic defeat mechanism. In conclusion, he provides two broad recommendations and three specific techniques as examples of the kind of innovation needed to out-disseminate opponents—at scale. Regardless of what course U.S. competitors pursue, Influence Machines will progress and proliferate because the attention trade is highly lucrative. Future IO systems, built by competitors and corporations alike, will be able to simultaneously monitor and affect tens of thousands of people at once, but the Pentagon’s current efforts to integrate AI lack sufficient IO functionality. The pieces are already there, waiting for an adversary to assemble its own Influence Machine, for which disinformation is simply a user setting.
The Defense Department has become fascinated with Artificial Intelligence (AI), and rightly so, as this technology will be as transformative as electricity was a century ago. This paper proposes that the convergence of AI and information operations (IO) represents a greater strategic liability than computationally similar applications—physical adaptations, like drone swarms, and informational applications, such as intelligence process optimization—that have drawn so much budgetary attention. Using a hypothetical “Influence Machine,” the article explores evolving techniques that achieved near-human acuity on many influence tasks. With all the necessary data essentially pre-structured, machine learning can perform these tasks at a massive scale. Using the historical parallel of the impact of the printing press during the Reformation, the author unpacks the exponential potential of emerging influence systems. He also examines how, applied during a time of crisis, such tools may provide a strategic defeat mechanism. In conclusion, he provides two broad recommendations and three specific techniques as examples of the kind of innovation needed to out-disseminate opponents—at scale. Regardless of what course U.S. competitors pursue, Influence Machines will progress and proliferate because the attention trade is highly lucrative. Future IO systems, built by competitors and corporations alike, will be able to simultaneously monitor and affect tens of thousands of people at once, but the Pentagon’s current efforts to integrate AI lack sufficient IO functionality. The pieces are already there, waiting for an adversary to assemble its own Influence Machine, for which disinformation is simply a user setting.
No comments:
Post a Comment