Subj : Re: ChatGPT?
To   : dozo
From : DustCouncil
Date : Sun Feb 19 2023 03:08:42

 do> The problem I'm having with it is that any content it generates is based
 do> on what it 'learned' or what it has been fed. The internet can be/is a
 do> cesspool which causes the generated content to be biased on what it has
 do> seen and learned. The problem with that is: who would police the AI so
 do> it's not coming off biased?

It's worse than that.  A form of terrorism in the future will be poisoning the data vacuumed up to feed these AIs.  

The future is coordinated smears against people they don't like posted on
blogs and social media, in the same way people do this now to mess with
search engine rankings.  In this case, it will be so that AI, having vacuumed
up this information to form sentences, will deliver the smear in the same way
as these AI experiments with chatbots go all Nazi, given enough time. 

I am not a Luddite, but the way we we are approaching AI right now, and I
would question whether this is AI in the way we understand that term, is
impetuous.  

What may stop it is a lawsuit when, at some point, a deployed AI gives
incorrect information that gets someone killed.  Imagine googling something
about drug interactions, and one of these landfill websites which generates
AI spew gives information that leads to someone's death.

I am less disturbed by ChatGPT than I am at what has been written about it.


https://www.scientificamerican.com/article/google-engineer-claims-ai-chatbot-i
s-sentient-why-that-matters/

People are losing their minds.

--- Mystic BBS v1.12 A47 2021/12/24 (Linux/64)
 * Origin: Shipwrecks & Shibboleths [San Francisco, CA - USA] (21:1/227)

.