Discussion about this post

User's avatar
Jojo's avatar

"Using AI-generated data to train AI could introduce further errors into already error-prone models. Large language models regularly present false information as fact. If they generate incorrect output that is itself used to train other AI models, the errors can be absorbed by those models and amplified over time….Even worse, there’s no simple fix."

---------

GIGO. The internet is littered with false and incorrect info. These days, anyone can publish virtually anything they want w/o anyone overseeing whether it is correct. If someone does correct the posting, it is typically done in a subsequent post that still leaves the original incorrect post up and available to be scraped by search engines.

Some months back I had some back and forth with the General Editor of the our small, free local newspaper. They had published a letter to the editor that was factually incorrect. I asked why the paper did not add a comment that while they were publishing the letter to show the opinion of a reader, the information was incorrect? The person told me that it wasn't their job to correct letters to the editor. Which makes them part of the problem.

Given this reality, why is anyone surprised that ChatGPT, et al produces bad answers, since it obtained its "knowledge" by sucking in everything on the web without curation?

Expand full comment
Jojo's avatar

"perfect music for enhanced focus"

For me, the perfect music for concentration is the sound of silence. The older I get, the more I prefer silence. I was very happy when my health club eliminated the din of background music, recognizing that everyone was wearing headphones or earbuds and wasn't listening to THEIR choice of music anyway!

Expand full comment
29 more comments...

No posts