The agency looks at a broad range of research, like ways to predict potential threats to national security and secure methods of communication, but the ones that are most ambitious have to do with artificial intelligence and the brain. Specifically, IARPA wants "Brain-inspired algorithms and architectures for artificial intelligence and machine learning." This kind of technology is already being explored by DARPA, who is funding IBM's Synapse project. They're also looking for an "autonomous agent, machine learning and/or physiological intelligence methods to gather latent or suppressed knowledge," which sounds a lot like an A.I. interrogator.

Here's a (condensed) list of some other things IARPA is looking for:

• Brain computer interfaces to enhance cognitive processing or increase bandwidth of human-machine interactions

• Computational social policy

• Reliable, real-time feedback methods for assessing human judgment and reasoning

• Methods for assessing capability and intent to develop weapons of mass destruction

• Methods for assessing capability and intent to leverage cyber capabilities against U.S. critical infrastructure

• Forensics on multimedia and/or social network data

• New algorithms and techniques that take advantage of quantum entanglement to perform tasks (excluding quantum annealing) that are inefficient or impossible with classical algorithms and/or current platforms

• Robotized research methods

• Tools to identify and mask signal streams and records that contain personal information to avoid unauthorized collection and dissemination

Now just because IARPA asks for these kinds of projects doesn't mean they're actually going to happen. It's more of a wish list than anything else, but gives insight into the kinds of technology the agency would like to employ.