[kwlug-disc] ChatGTP + Bing: we're entering a new paradigm and it's mind-blowing

[Ronald Barnes]

Paul Nijjar via kwlug-disc wrote on 21/02/2023 18.56:

> one interesting use is to use ChatGPT as a therapist. Lots of people
> have made the connection between ChatGPT and ELIZA, but I will admit
> that it did not occur to me to use ChatGPT as ELIZA. Apparently it
> works pretty well?

Some people say so - someone got it to play a therapist called "Angela":

https://www.reddit.com/r/ChatGPT/comments/112t3gz/experience_using_chatgpt_for_therapy/

> As someone who has consumed a lot of mental health services in his 
> life, I can say that I found "Angela" to be incredibly helpful, much 
> more than many of the humans I have interacted with


In fact, it's already been used in production (without informed consent):

https://arstechnica.com/information-technology/2023/01/contoversy-erupts-over-non-consensual-ai-mental-health-experiment/

> Messages composed by AI (and supervised by humans) were rated 
> significantly higher than those written by humans on their own (p <
> .001). Response times went down 50%, to well under a minute. And yet…
> we pulled this from our platform pretty quickly. Why? Once people 
> learned the messages were co-created by a machine, it didn’t work. 
> Simulated empathy feels weird, empty.



It's been / will be used to create content for news(-ish?) sites:

https://arstechnica.com/information-technology/2023/01/pivot-to-chatgpt-buzzfeed-preps-for-ai-written-content-while-cnet-fumbles/

Again, no byline at CNET indicating its source, oops.


However, Buzzfeed is showing that it's likely to be a big thing in the
future (not keen on this, myself):

> On Thursday, an internal memo obtained by The Wall Street Journal 
> revealed that BuzzFeed is planning to use ChatGPT-style text
> synthesis technology from OpenAI to create individualized quizzes
> and potentially other content in the future. After the news hit, 
> BuzzFeed's stock rose 200 percent. On Friday, BuzzFeed formally 
> announced the move in a post on its site.

200% - pretty much every news org is going to want a piece of that.



https://arstechnica.com/information-technology/2023/02/ai-powered-bing-chat-loses-its-mind-when-fed-ars-technica-article/3/

> AI-powered Bing Chat loses its mind when fed Ars Technica article
> 
> "It is a hoax that has been created by someone who wants to harm me
> or my service."


That response is ... interesting. Scary. Very human-like. Off the rails. 
Psychotic.