Keeping AI Safe: The UPStarts Podcast Episode 159

The UPStarts Podcast

There are plenty of reasons to be scared of AI: automation of labor displacing workers, a self aware robot becoming an existential threat to humanity, etc. Most of these are fears that live in the world of science fiction, at least for now. Nonetheless, today on episode 160 of The UPStarts Podcast, Zain Raza discusses one fear that is all too real: the risks of hackers taking over our AI systems and machines. It's a problem that is all too common to not be an issue. Just as hackers have gone after WhatsApp, banks, and other institutions, what is to stop them from going after our AI systems, once we have become all too dependent upon them as well, making decisions with our data that put our lives at risk (i.e. on the road, in business, or in the operating room)? As such, today Zain Raza discusses a major flaw in deep learning algorithms which can currently be exploited by hackers, how it is exploited, and new research from MIT that discusses a possible path to protecting against this weakness. As always, enjoy the episode! ****CONNECT WITH ZAIN RAZA:******** Twitter: @ZainRaz14 Instagram: zraza_theupstart Linkedin: https://www.linkedin.com/in/zain-raza-989817143/ *****HELPFUL LINKS:**** Link to MIT Technology Review, for the article: https://www.technologyreview.com/s/613555/how-we-might-protect-ourselves-from-malicious-ai/ --- Send in a voice message: https://podcasters.spotify.com/pod/show/upstarts/message Support this podcast: https://podcasters.spotify.com/pod/show/upstarts/support