NOVEMBER 6, 2019 – A next war against a near-peer competitor will be fast, chaotic and shockingly bloody, the director of the Joint Artificial Intelligence Center said.
Air Force Lt. Gen. Jack Shanahan spoke at the National Security Commission on Artificial Intelligence today in Washington.
The side with the best AI algorithms will put the other side at an extreme disadvantage, he said, particularly regarding the speed of battlefield decision making.
Shanahan described such a future battle as “algorithms vs. algorithms,” with the best algorithm victorious.
In the future battle scenario, Shanahan said events will move so quickly that a traditional chain of command won’t work. Junior, frontline troops will need to be empowered to make the decisions and to adjust AI algorithms on the fly.
This decentralization of command entails higher risks and consequences, he said, but without it, “we risk losing the fight.”
To get to the best AI, the department must rely on industry and academia, which are much further along in this endeavor than the DOD, he said.
Shanahan said there are lessons learned from Google’s unwillingness to continue working with the DOD on Project Maven last year. The project had to do with AI’s use in intelligence, surveillance and reconnaissance operations.
There needs to be a shared sense of responsibility and vision, along with trust and transparency from the DOD and industry, he said. “National security depends on it.”
Another step to take in adopting the best of industry’s AI is for service members to work directly in industry and academia and to bring AI experts from industry and academia into the DOD. “That’s already happening but we need to scale that up,” he said. “Peer-to-peer discussions and personal relationships matter.”
Lastly, Shanahan said an important step that was taken last week was the Defense Innovation Board coming up with a set of AI ethical principles, which he said are excellent.
The DOD will act on the Defense Innovation Board’s recommendations and then there will be deliberations on implementing them if the recommendations are accepted. “Implementation is not an overnight task,” he said.
The ethical use of AI by the military in training, research, product development and operations should inspire confidence in the industry that the department is making ethical use of this new technology.
“China and Russia didn’t hold public hearings on the ethical use of their AI and I never expect them to do this,” he added.
BY DAVID VERGUN