FIRST AIRED: June 5, 2018

Nice work! Enjoy the show!


You’re busy. We get it.

Stay on top of the news with our Editor’s Picks newsletter.

US Edition
Intl. Edition
Unsubscribe at any time. One click, it’s gone.

Thanks for signing up!



>> The Pentagon is hoping to tap the emerging field of artificial intelligence for the daunting task of tracking an enemy's nuclear missiles. Officials tell Reuters that the secret project could give the United States an edge as it confronts nuclear threats that now include North Korea. Pentagon correspondent Phil Stewart is on this story.
>> So right now, the US Air Force, the US Military through satellites and other manners of collecting intelligence have an enormous amount of data. And what advocates of using artifical intelligence say is that this technology is gonna allow them to sift through all of this data, and find the needle in the haystack.
And the needle in the haystack in this case is basically signs that a missile launched might be imminent. And that would give the US military potentially lot more lead time that it has right now to prepare for a launch, and advice policy makers who would have make big decisions about whether to attack and strike those missiles.
>> The idea would be to deliver a rapid AI assisted alert to US commanders even before an adversary's missile takes flight. Officials believe Russia and China are actively pursuing similar technologies, and the US must catch up. But others warn of serious risks.>> Current research on artificial intelligence shows that yes, those systems can be very easily spoofed.
There was a great study done by MIT that showed that a turtle if manipulated the right way could show up in the Google image processor as a rifle. So, what if the North Koreans were able to manipulate their missiles in a way that maybe United States think these were other kinds of objects, maybe a car or a truck, and not actually missiles?
What if they made things that were cars and trucks look like missiles just to spoof the United States? What if China or Russia did something similar? So the idea of having computers make decisions based on imagery and other intelligence raises serious questions about creating false alerts and false alarms.
And whether or not those could lead the United States down to decision making process that in itself could lead to war.>> The Pentagon is moving cautiously, with AI assisted missile defense possibly coming online in the early to mid 2020s.