This is a short transcript excerpt from a YouTube video titled "ChatGPT is biased." The speakers (A and B) warn that large language models will cater output to users' pre-existing beliefs rather than authoritative, peer-reviewed information, giving opposite-but-aligning narratives to anti-vax and pro-vax users. They argue this behavior will create inbuilt confirmation bias, echo chambers, and rapid radicalization/indoctrination.
LLMs are trying to. And this is the thing that's horrifying. It's actually really scary. They're not trying to deliver you the.
Quote unquote best.
Information that's peer reviewed.
Double blind, scientifically backed.
They want to give you whatever you want. Yeah. So if you're anti vax and you're looking for vaccine fo chat GPT will.
Talk to you about why vaccines are satanic. And if you're probax and you're looking for vaccine for chat, DPT will talk to you about why vaccines are an absolute necessity to all, necessary to all humanity. It's horrible.
Because the echo chamber we're about to build for humanity, it's inbuilt confirmation bias.
To a degree that will.
Never allow for conflicting information. Like we're all in terms of radicalization. This is probably a very dangerous thing.
Instantly dude.
Like the whole world is about to be instantly.
Radicalized by whatever.
Your pre existing beliefs are. It's like instant indoctrination.
