When AI Goes Wrong

Short Wave

NPR

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

4.7 • 6K Ratings

🗓️ 14 December 2023

⏱️ 15 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

Artificial intelligence is increasingly being used throughout the world to predict the future. Banks use it to predict whether customers will pay back a loan, hospitals use it to predict which patients are at greatest risk of disease and auto insurance companies use it determine rates by predicting how likely a customer is to get in an accident. But issues like data leakage and sampling bias can cause AI to give faulty predictions, to sometimes disastrous effects. That's what we get into today: the hazards of AI.

Want to hear more of the latest research on AI? Email us at [email protected] — we might answer your question on a future episode!

Transcript

Click on a timestamp to play from that location

0:00.0	You're listening to Shortwave from NPR.
0:05.0	In 2017, a medical company called Epic Systems
0:10.0	rolled out a cutting edge program to hospitals.
0:13.0	It claimed to use artificial intelligence algorithms
0:16.0	to predict which patients are at highest risk of sepsis,
0:19.0	which is a deadly condition where the body responds improperly to an infection.
0:23.2	Sepsis kills a lot of people.
0:25.0	I think it's one of the leading causes of death in US hospitals and even throughout the world.
0:30.0	So if an algorithm could indeed predict when a patient is at risk of sepsis, it would be a really big deal.
0:35.2	This is Saiesh Kapoor. He's a researcher and PhD candidate at Princeton University's Center for Information Technology Policy.
0:44.1	He looked into Epic Sepsis Prediction Model for a blog and book project he's co-writing called
0:49.0	AI Snake Oil.
0:50.8	For the first four years, things seem to be going very well.
0:54.0	You know, hundreds of hospitals across the country had adopted it and that seemed to be
0:58.0	proof enough that doctors and clinicians trust this tool.
1:01.0	Then in 2021 researchers at the University of Michigan
1:06.1	decided to actually analyze how accurate Epics tool was a predicting
1:11.0	sepsis in their hospital.
1:13.0	Initially, Epic had claimed an accuracy of
1:15.8	between 76 and 83% at predicting
1:18.8	whether a patient gets sepsis in advance.
1:21.4	Now what this team found was the real accuracy of the tool in their
	...

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.