That Thread Written by ChatGPT

[Belacqua]

Real me...

I simply asked ChatGPT to write a post for atheistforums.org from the perspective of Swedenborg. Then I asked it to ridicule the first first post.  For the reply to Fake Messiah, I asked for reply that explains Swedenborg's belief in life on other planets in terms of alchemical symbolism. I really did not have any overt intentions by creating the thread other than being transparent by 1) posting in "Off Topic", 2) putting ChatGPT in the title, and 3) posting the uneditted text.
Kinda just wanted to see what would happen. Perhaps some people would share their own generated text for replies. Or their own experiences with ChatGPT.

@Bel, there is a desktop version that does not require a phone.

@Fake, the "cut and paste" rule IMHO would not apply since the generated text was prompted by my own seed text and not publicly available. That said, I could see a rule against entirely AI generated text, since that undermines the social aspect of social media although how the mods would detect and enforce such a rule IDK.


One thing I have noticed is that ChatGPT tends to be very wordy.

@emjay, ChatGPT interpreted my initial request to produce an atheistic take on Swedenborg. I had asked for a post for atheist forums from a Swedenborgian perspective. It assumed I was an atheist. 

It's amazing to me that it could generate such clear, coherent texts, with impressive detail, from such simple prompts. I actually don't know enough about Swedenborg to know if what it said is correct. One article I read, after seeing the OP, said that AI is good at appearing confident about asserting things that are wrong. 

My first reaction is that this is going to take over certain practical uses for writing text. No doubt there is, or soon will be, a version in which a researcher can input a series of facts and the AI will provide him with a paper in a publishable format. Papers reporting research results published in medical journals, which discourage creativity and demand adherence to formula, seem very suitable to automatic generation. And why not? As long as the researcher checks the results, why not get the computer to do the labor of writing it out? 

Maybe we can make an analogy with photography. Cheaply reproduced photos, as you know, took over certain jobs from artists. Picture postcards replaced engravings as souvenirs for tourists. Really good engravings are far superior, aesthetically, to picture postcards, but the easier option is more affordable. 

When I use G-mail the computer brain often suggests how I should finish the sentence I'm writing. No doubt this would be useful if I were sending business letters, but I am always careful NOT to use the language the computer suggests. If we want to have any pretense at a prose style, or original thought, that seems important to me. 

Likewise I think of stylists like Nabokov, who have such a horror of cliché. He never ever wrote a sentence that contained the predictable expression. I do not think that an AI is going to approach his level of wit, multi-level irony, and subtle allusion -- not for a very long time. So maybe that's an analogy to utilitarian photos vs. creative painting. 

Also curious is how the ChatGPT has better manners than either Fake or Nudge, even when it was asked to ridicule.

This makes me wonder how much the AI follows guidelines installed by its programmers. Surely it could be programmed to be nasty...? Although effective nastiness requires knowing what (you think) would be personally hurtful to the reader. 

Anyway, it's a fascinating experiment. I'm glad you did it.