What to know about an AI transcription tool that ‘hallucinates’ medical interactions
PBS News Hour - Segments
PBS NewsHour
4.1 • 1K Ratings
🗓️ 25 January 2025
⏱️ 6 minutes
🧾️ Download transcript
Summary
Transcript
Click on a timestamp to play from that location
| 0:00.0 | As artificial intelligence increasingly becomes part of daily life, both its benefits and its |
| 0:06.1 | pitfalls are becoming apparent. Take medical centers. Many of them use an AI-powered tool called |
| 0:12.1 | whisper to transcribe patients' interactions with their doctors. But researchers have found that it |
| 0:17.8 | sometimes invents text. It's what's known in the industry as hallucinations. |
| 0:22.9 | That raises the possibility of errors like misdiagnosis. |
| 0:27.0 | Garantz Burke is an Associated Press Global Investigative reporter who's been looking into this. |
| 0:32.1 | I first want to give folks an example of what researchers found. |
| 0:40.3 | Here's what a speaker said. And after she got the telephone, he began to pray. Simple sentence. But here's what was transcribed. |
| 0:46.3 | Then he would, in addition to make sure I didn't catch a cold, he would help me get my shirt, kill me, and I was, he began to pray. |
| 0:57.2 | What sorts of other hallucinations have been found? |
| 1:01.3 | Yeah, so in talking with more than a dozen engineers and academic researchers, |
| 1:07.6 | my co-reporter Hilke Shalman and I found that this particular AI-powered transcription |
| 1:13.6 | tool makes things up that can include racial commentary, sometimes even violent rhetoric. And, of course, what we're talking about here, you know, incorrect words regarding medical diagnoses. So that office leads to a lot of |
| 1:32.0 | concerns about its use, particularly in really sensitive settings like in hospitals. |
| 1:37.9 | We asked OpenAI about this, and here's what they told us. They said, we take this issue |
| 1:42.1 | seriously and are continually working to improve the |
| 1:45.4 | accuracy of our models, including reducing hallucinations. For Whisper, our usage policies |
| 1:51.2 | prohibit use in certain high-stakes decision-making contexts, and our model card for open-source use |
| 1:57.6 | includes recommendations against use in high-risk domains. Given those warnings, |
| 2:03.3 | why does so many medical centers use this? You know, I think we're at a time when a lot of |
| 2:08.6 | healthcare systems are looking to AI and AI agents to do things that human beings do more efficiently |
| 2:16.5 | and at scale. There's a lot of talk about the promise |
... |
Please login to see the full transcript.
Disclaimer: The podcast and artwork embedded on this page are from PBS NewsHour, and are the property of its owner and not affiliated with or endorsed by Tapesearch.
Generated transcripts are the property of PBS NewsHour and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.
Copyright © Tapesearch 2026.

