A pair of think-tanks are pushing a combination of machine learning and open-source data as a way to generate more leads on potential nuclear proliferation and said Tuesday that a recent pilot project shows the technique “could save hundreds of analyst hours” if put into practice by the government.
That was the claim in a Zoom even hotels by the Washington-based Nuclear Threat Institute and Center for Advanced Defense Studies, which invited the public to sit for a presentation about their conclusions in a January report on the project, “Signals in the Noise: Preventing Nuclear Proliferation with Machine Learning & Publicly Available Information.”
A couple of heavy-hitting non-proliferation pros affiliated with the Nuclear Threat Institute, former Sen. Sam Nunn (D-Ga.) and former Secretary of Energy Ernest Moniz, said the sponsors should take their machine-learning tool to the Biden administration and the international community.
Moniz, now the institute’s co-chair and chief executive officer, suggested that the incoming Biden administration, and Biden personally, might be receptive to the idea of leveraging machine-learning to track proliferators.
“Joe Biden, when he was [vice president under Barack Obama] in charge of the cancer moonshot, what we at [the Department of Energy] took to him was exactly these ideas: that mining and machine learning in vast amounts of heterogeneous cancer data might be the answer to lots of challenges.” Moniz said.
“How do you correlate this public information data and all the machine learning and the things you’re going to do with the intelligence close to us? And where is that going to get done?” Nunn asked. “There’s a whole set of questions about legitimacy that may lend itself or may require an international organization for credibility’s sake,” Nunn said.
