US military to build AI that predicts adversaries’ next moves

One of the essential aspects of the program is not only to understand the current strategies of an adversary but also how those strategies will evolve under different conditions.

Theory of Mind, a new initiative by the US Defence Advanced Research Projects Agency (DARPA), which aims to enhance national security and decision-making, is under the works.

The project seeks to develop new technologies in order to enable decision makers to get a better understanding of potential adversaries, engage with them and develop strategies for deterrence and incentive-based actions.

Theory of Mind program will combine advanced algorithms with human expertise, DARPA announced, as per Interesting Engineering.

A high-level modelling and simulation environment will be used to look at various national security scenarios. This will provide decision makers with an array of options and improve efficiency in assessing potential courses of action.

Theory of mind refers to the ability of humans to infer and understand the beliefs, desires, and intentions of others. Artificial Intelligence, since its inception has attempted to replicate various human behaviors in computational form to a point where 'machine behavior' is indistinguishable from human activity, even in the same conditions.

One of the essential aspects of the program is not only to understand the current strategies of adversaries but also how those strategies will evolve under different conditions.

Eric David, a recent recruit of the US Department of Defense (DoD), DARPA, who joined earlier this year in February 2023 as the lead scientist for this initiative.

Davis prior to this specialised in artificial intelligence, machine learning , and human-machines collaboration at Galois. Galois is a research and development firm that has worked with a number of government organisations, including the US Intelligence Community and NASA.

The announcement by DARPA did not specify which adversaries however the implication of developing such an algorithm raises concerns. A major concern with the use of such technology is how it might be used.

DARPA has previously engaged in efforts to monitor, predict and influence human behaviour by collecting and analysing large amounts of data.

Following the September 11, 2002 attacks, DARPA launched a program called Total Information Awareness (TIA). However this is initiative is no longer in use.

In order to enhance US' capacity to detect and preempt terrorist activities by strengthening surveillance capabilities, TIA was launched. Many, including the American Civil Liberties, criticised the TIA a significant overreach.

Certain aspects of the Theory of the Mind bares similarities with DARPA previous initiatives. TIA involved developing data-mining tools to screen through copious amounts of information to identify patterns related to potential threats. Theory of Mind similarly, aims to provide insights into behaviour of adversaries. However, information about the methodology and the technological aspects is unknown.

DARPA's LifeLog program which was announced in 2003, and abandoned shortly after, aimed to create a comprehensive database of individual experience and perform as a personal assistant.

Its ultimate objective, which aligns with the goals of the Theory of Mind, was to assist people in properly comprehending and managing their experiences and preferences.

However DARPA's Theory of Mind initiative raises ethical concerns about developing algorithms to interpret human behaviour and increased scrutiny from civil rights advocates.