Dr. Catherine Marsh, director of the Intelligence Advanced Research Projects Activity (IARPA) part of the Office of the Director of National Intelligence, talked about the value – in limited doses – of failure as an expected and necessary outcome of the difficult research areas that her organization was created to take on with the help of industry and academia.
Speaking on April 11 at an event organized by the Intelligence and National Security Alliance (INSA), Marsh said that IARPA undertakes research into “disruptive, next-generation technologies” that have the promise to have a high pay-off rate. At the same time, the organization also builds a significant failure rate into that equation.
“I’d like to say that we have an unbroken record of success, but that’s just not the case,” she explained, adding that failing to break through and discern the next great application of technology for the intelligence community is part of the game “if you take a lot of risks.”
“If you don’t fail, it means you aren’t taking enough risks,” Marsh counseled. “Failure happens,” she said. “It’s vital to our work – as long as you learn something from it.”
She also talked about her organization’s mission, which she said aims to support all 18 entities within the U.S. intelligence community, and which leads to a high degree of collaboration across agencies.
Marsh ticked off numerous project areas in that IARPA is working, among them quantum science, psychology-driven subjects, and biological issues. Woven thoroughly through the project list is the use of artificial intelligence and machine learning to harness and direct data discoveries, with an emphasis for some projects on image data generated by remote sensing devices, and internet of things-generated data.
Part of the organization’s work, she said, is to “navigate a sea of data” and try to make it useful to decision-makers. Within that equation, she explained, is a constant evaluation of the variety, value, velocity, and veracity of data, which is continually changing.
The IARPA director also listed several projects that the agency is either seeking help from industry and academia right now or is expected to announce in the near term. These include examinations of:
- AI-based technologies to attribute authorship;
- Technologies to create micro-simulations of human movements; and
- How to create navigational models for use in very remote environments for which such data is not readily available.