AI Cognitohazard (E372)

QAA Podcast

Julian Feeld, Travis View & Jake Rockatansky

News

4.3 • 4.6K Ratings

🗓️ 13 May 2026

⏱️ 66 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

In 2025, researchers at the University of California submitted an academic article related to a case study of what they called “New-onset AI-associated Psychosis”. The article breaks down the case of a 26 year old woman who was hospitalized after she began to believe her brother was speaking to her through ChatGPT. Unfortunately, this was not an isolated incident and in today’s episode Liv examines similar stories that have emerged in the wake of the AI chatbot surge. Along the way, she explores how the companies behind these chatbots plan to prevent situations like this from happening - or if they plan to at all - and even shows what happens when a user becomes overly reliant on their AI and begins to sound like one themselves. In Jake’s words: “They are not making anything that’s dope.” Subscribe for $5 a month to get all the premium episodes: www.patreon.com/qaa Check out our new podcast series network Cursed Media! All episodes of Spectral Voyager Season 2 are out now! Binge the entirety of Truly Tradly Deeply by Annie Kelly and Megan Kelly as well as Science in Transition by Liv Agar and Spencer Barrows: cursedmedia.net Produced by Liv Agar & Corey Klotz. Theme by Nick Sena. Additional music by Pontus Berghe. Theme Vocals by THEY/LIVE (instagram.com/theyylivve / sptfy.com/QrDm). Cover Art by Pedro Correa: (pedrocorrea.com) qaapodcast.com QAA was known as the QAnon Anonymous podcast.

Transcript

Click on a timestamp to play from that location

0:00.0	I'm
0:02.0	Oh,
0:03.0	Oh,
0:04.0	Oh,
0:05.0	Oh,
0:07.0	Oh,
0:08.0	Oh, If you're hearing this, well done, you found a way to connect to the internet.
0:36.2	Welcome to the QAA podcast, episode 371, AI Cognito Hazard.
0:43.2	As always, we are your host, Jake Rakitansky.
0:45.5	Liv Egar.
0:46.5	Julian Field.
0:47.5	And Travis View.
0:48.9	In the 1960s, a computer scientist at MIT named Joseph Weisenbaum attempted to create an automated chatbot
0:55.5	that was capable of having conversations with a human being.
0:58.5	He would call the bot Eliza, named so after the character Eliza Doolittle from Bernard Shaw's Pygmalion.
1:05.0	Just as Doolittle slowly accumulated more knowledge about how to speak with an upper-class inflection,
1:09.4	so too did Weisenbaum's Eliza slowly learn how to better respond to its carbon-based patients.
1:14.6	In order to skirt around the problem of being more than half a decade too early for
1:17.9	sufficiently robust large language models, Weisenbaum would base Eliza around the speech patterns
1:22.8	of a psychotherapist, typically answering prompts by rephrasing them in the form of a question.
1:27.3	Here's a real chat log I pulled from a CBC article.
1:30.3	Please tell me your promulmblem.
	...

Transcript will be available on the free plan in 6 days. Upgrade to see the full transcript now.

Previous episode | Next episode

Disclaimer: The podcast and artwork embedded on this page are from Julian Feeld, Travis View & Jake Rockatansky, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of Julian Feeld, Travis View & Jake Rockatansky and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.