![\"A](\"https:\/\/cdn.defenseone.com\/media\/img\/upload\/2018\/03\/16\/AP_916483777771\/defense-large.jpg\")
From eastern Europe to southern Iraq, the\u00a0U.S.<\/span>\u00a0military faces a \u00a0difficult problem: Adversaries pretending to be something they\u2019re not \u2014 think Russia\u2019s \u201clittle green men\u201d in Ukraine. \u00a0But a new program from the Defense Advanced Research Projects Agency seeks to apply artificial intelligence to detect and understand how adversaries are using sneaky tactics to create chaos, undermine governments, spread foreign influence and sow\u00a0discord.<\/p>\nThis activity, hostile action that falls short of \u2014 but often precedes \u2014 violence, is sometimes referred to as gray zone warfare, the \u2018zone\u2019 being a sort of liminal state in between peace and war. The actors that work in it are difficult to identify and their aims hard to predict, by\u00a0design.<\/p>\n
\u201cWe\u2019re looking at the problem from two perspectives: Trying to determine what the adversary is trying to do, his intent; and once we understand that or have a better understanding of it, then identify how he\u2019s going to carry out his plans \u2014 what the timing will be, and what actors will be used,\u201d said\u00a0DARPA<\/span>\u00a0program manager Fotis\u00a0Barlos.<\/p>\n Dubbed\u00a0COMPASS<\/span>, the new program will \u201cleverage advanced artificial intelligence technologies, game theory, and modeling and estimation to both identify stimuli that yield the most information about an adversary\u2019s intentions, and provide decision makers high-fidelity intelligence on how to respond\u2013-with positive and negative tradeoffs for each course of action,\u201d according to a\u00a0DARPA<\/span>\u00a0notice posted Wednesday.<\/p>\n Teaching software to understand and interpret human intention \u2014 a task sometimes called \u201cplan recognition\u201d \u2014 has been a subject of scholarship since at least\u00a0a 1978 paper\u00a0by Rutgers University researchers who sought to understand whether computer programs might be able to anticipate human intentions within rule-based environments like\u00a0chess.<\/p>\n Since then, the science of plan recognition has advanced as quickly as the spread of computers and the internet, because all three are intimately\u00a0linked.<\/p>\n From Amazon to Google to Facebook, the world\u2019s top tech companies are pouring money into probabilistic modeling of user behavior, as part of a constant race to keep from losing users to sites that can better predict what they want. A user\u2019s every click, \u201clike,\u201d and even period of inactivity adds to the companies\u2019 almost unimaginably large sets, and new machine learning and statistical techniques (especially involving\u00a0Bayesian reasoning) make it easier than ever to use the information to predict what a given user will do next on a given site. (Among these tools is\u00a0Google\u2019s Activity Recognition library,\u00a0which helps app developers imbue their software with a\u00a0better sense of what the user is\u00a0doing.)<\/p>\n But inferring a user\u2019s next Amazon purchase (based on data that user has volunteered about previous choices, likes, etc.) is altogether different from predicting how an adversary intends to engage in political or unconventional warfare. So the\u00a0COMPASS<\/span>program seeks to use video, text, and other pieces of intelligence that are a lot harder to get than shopping-cart\u00a0data.<\/p>\n The program aligns well with the needs of the Special Operations Forces community in particular. Gen. Raymond \u201cTony\u201d Thomas, the head of\u00a0U.S.<\/span>\u00a0Special Operations Command, has said that he\u2019s interested in deploying forces to places before there\u2019s a war to fight. Thomas\u00a0has discussed\u00a0his desire to apply artificial intelligence, including neural nets and deep learning techniques, to get \u201cleft of\u00a0bang.\u201d<\/p>\n Unlike shopping, the analytical tricks that apply to one gray-zone adversary won\u2019t work on another. \u201cHistory has shown that no two [unconventional warfare] situations or solutions are identical, thus rendering cookie-cutter responses not only meaningless but also often counterproductive,\u201d wrote Gen. Joseph Votel, who leads\u00a0U.S.<\/span>\u00a0Central Command, in\u00a0his seminal 2016 treatise\u00a0on gray zone\u00a0warfare.<\/p>\n As practiced by Amazon and others within the domain of online shopping, \u201cplan recognition\u201d at scale is very cookie-cutter. If\u00a0COMPASS<\/span>\u00a0succeeds, it will have to apply game theory and big data to behavior prediction in ways that Silicon Valley has never attempted. And it will have to do so repeatedly, in the face of varied and constantly morphing adversaries looking to keep as much of their activity hidden as possible \u2026 an ambitious undertaking even for\u00a0DARPA<\/span>.<\/p>\n