Victoria Coleman, Director of the Defense Advanced Research Projects Agency (DARPA) and her deputy Peter Highnam, discuss the agency’s past successes as well as future projects with columnist David Ignatius. Coleman explained that DARPA is pressing forward despite the pandemic and still looking for what she describes as “very risky, big bets, ‘if it works it changes everything’ kinds of ideas.”
DARPA for many years has funded breakthrough research in quantum computing. The agency in September announced a program to use room-temperature atomic “vapors” for quantum sensing and imaging. (The acronym-addicted agency calls this one SAVaNT, for Science of Atomic Vapors for New Technologies.) Meanwhile, another DARPA program is combining “noisy” (meaning short-lived) qubits with classical computers to solve some otherwise impossible optimization problems.
Ignatius includes coverage of several advanced technologies under DARPA auspices. DARPA’s program called Lifelong Learning Machines, started three years ago, is trying to build true “learning machines” on the premise that “even the smartest of the current crop of AI [artificial intelligence] systems can’t stack up against adaptive biological intelligence.”
Artificial intelligence has been a DARPA project for several decades, but the pace is accelerating. Coleman told me the agency is spending $2 billion over the next five years to fund more than 60 AI programs, including more than 30 that are exploring “next generation” AI.