Adversarial AI

Short Wave

NPR

Daily News, Nature, Life Sciences, Astronomy, Science, News

4.7 • 6K Ratings

🗓️ 24 October 2019

⏱️ 10 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

Artificial intelligence might not be as smart as we think. University and military researchers are studying how attackers could hack into AI systems by exploiting how these systems learn. It's known as "adversarial AI." Some of their experiments use seemingly simple techniques. Dina Temple-Raston has been looking into this for her special series, I'll Be Seeing You. Follow host Maddie Sofia on Twitter: @maddie_sofia. Or email the show at [email protected].

Transcript

Click on a timestamp to play from that location

0:00.0	You're listening to Shortwave from NPR.
0:04.6	Hey everybody, Maddie Sifaya here again.
0:08.1	This time with NPR Special Correspondent, Dina Temple-Raston.
0:11.2	Hey Dina.
0:12.2	Hey there.
0:13.2	So you're here because you've been doing some really cool reporting about artificial
0:16.8	intelligence as part of your special series.
0:19.4	I'll be seeing you.
0:20.4	Yeah, we did a story that it was explaining how AI works and how it's finding its way
0:25.2	into everything from refrigerators to insurance even conservation.
0:29.6	Like you also found out that for all of its potential, there are some real concerns about
0:34.1	hacking into AI.
0:35.1	There's actually a whole field of study that is focused on this.
0:39.0	It's called adversarial or evil AI.
0:41.8	Evil AI.
0:42.8	And it's a big enough concern that DARPA, the military's research arm, has created this
0:48.0	whole program to study it.
0:49.6	And it's called Guaranteeing AI robustness against deception.
0:54.4	Or luckily it has a short name guard.
0:56.8	The government is so good at naming things, Dina.
0:58.9	It is quite the name.
0:59.9	So DARPA is really good at creating tongue twisters.
	...

Please login to see the full transcript.

Previous episode | Next episode

Disclaimer: The podcast and artwork embedded on this page are from NPR, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of NPR and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.