ChatGPT & MidJourney: Useful Tool or Potential Handicap?

[Alatariel]

yea totally, it's a tool and a greatly beneficial one. But it's also being abused and that's the issue. I'll happily still use it for brainstorming purposes.

[Valhalla Erikson]

My concern is that, after so much misuse, programmers will tweak ChatGPT to the point that its no longer fun to use.

[Alatariel]

Mar 5, 2023 13:21:06 GMT -6 Valhalla Erikson said:

My concern is that, after so much misuse, programmers will tweak ChatGPT to the point that its no longer fun to use.

Definitely a valid concern. 

[Valhalla Erikson]

So to anyone using it my advice is to cherish your use while you can. But don't abuse it.

[Deleted]

Well, let me know if/when we want to see what the AI can do. We don't necessarily need a whole forum for it, just a disclaimer in the title that it was AI generated.

[Deleted]

Well, I wrote an entire story with ChatGPT. It's about 8,000 words, though I could probably condense it to under 5,000 if I really tried. The plot, characters and everything was all ChatGPT. I merely stated that I wanted to write a horror short story, and asked it to make a plot, and it did, and I went with it. I did do an edit afterwards for clarity, to fix a few issues, and maybe added a paragraph or two, but over 90% of the story is ChatGPT.

And...I'll let you decide how it is. If you wanna read it, I'll send you the link to the Google Doc. I'm not sure if posting links here is allowed. I am not planning on submitting this to any publishers or using it for any contests. I just wanted to see what ChatGPT would come up with if I let it play.

[Valhalla Erikson]

Mar 12, 2023 19:15:27 GMT -6 @tglassy said:

Well, I wrote an entire story with ChatGPT. It's about 8,000 words, though I could probably condense it to under 5,000 if I really tried. The plot, characters and everything was all ChatGPT. I merely stated that I wanted to write a horror short story, and asked it to make a plot, and it did, and I went with it. I did do an edit afterwards for clarity, to fix a few issues, and maybe added a paragraph or two, but over 90% of the story is ChatGPT.

And...I'll let you decide how it is. If you wanna read it, I'll send you the link to the Google Doc. I'm not sure if posting links here is allowed. I am not planning on submitting this to any publishers or using it for any contests. I just wanted to see what ChatGPT would come up with if I let it play.

It is allowed. I've been wanting to use it for a short story. Mostly outlining where I want my story to go. What prompt did you use for it to produce 8k words?

[Deleted]

Lots of them. First, I worked out the story, though ChatGPT did most of the work for this particular piece. Normally, it's all my story and I just use ChatGPT to help brainstorm, but this time I let it do what it wanted, for the most part. Then I'd write something to the effect of "Write the opening scene." And explain what I wanted out of that opening scene, the style I wanted and any important mmoments. I had to constatly go back and redo the prompt, reminding it to put descriptions in, keep it a close pov and add dialogue, otherwise it would default to just telling a summary of events. It would go for a scene or half a scene, then I would prompt it to continue with what I wanted out of the next scene. Rinse, repeat. It does really well with "write a Steven King style nightmare".

Once I was done, I took the whole thing and put it in a Google Doc and edited for clarity, adding in a couple things I wanted that ChatGPT didn't do or couldn't seem to grasp, but that all amounted to a couple paragraphs here and there and a few corrected sentences. Otherwise, it's a good 90% or more ChatGPT's writing. Yeah, I directed it, but I directed based off the outline ChatGPT made in my first few prompts in the chat.

Here's the link: docs.google.com/document/d/1wucdEIymC6lfaLQ9pL64L0st5kw3auCYIrpQAjxsD4I/edit?usp=sharing

Is it good? Meh. It's servicable. But remember, the plot was all ChatGPT. If I'd had my own plot, perhaps things would have been different. Though, I still feel that in my own writing, I'll only use it here and there to spice up a description of a scene or whatnot.

However, I had it write a short...well, essay, monologue, whatever, from the perspective of a character I've been working on for a while, named Conroy Quinn, and that turned out great. Quinn is a Listener, which means he can read minds. I told ChatGPT who he was, what he was like, his personality, then asked it to describe Listening from the perspective of Conroy Quinn. I added that at the end of the link above. It's only a couple paragraphs, but it captured Quinn's personality pretty well, I thouhgt.

[Valhalla Erikson]

I never understood why it doesn't want to do dialogue. I remember I wanted it to do a short story and it took me a lot of redos before I had it eventually give me the dialogue I wanted.

[ScintillaMyntan]

Welp, Submission Grinder just added a listing for a publication that is specifically looking for entirely machine-generated stories. thegrinder.diabolicalplots.com/Market.aspx?mid=15361

The weirder and less mainstream, the better. Bonus points if you can prompt your writing partner to write a Sword and Sorcery for the ages. We’d also love to see if you can get your new friend to open up and share a laugh.

This all makes me so uncomfortable. First I pride myself on writing weird stories. I don't want that to be devalued in the long run if people assume my unconventional ideas came from AI and not from my own very individual experiences, psychology, and creativity. People come up with the stories they do for very human reasons, even when they seem arbitrary. And then, a bunch of algorithms trained to respond to what you say based on patterns isn't your 'friend.'

In middle school I read "I, Robot" by Asimov and found the chapter where the little girl's only friend is a robot really depressing. At that point, the most I'd 'interacted' with a computer was when I tried to close a game and a box popped up saying "Are you sure you want to quit?" I could've played for hours and hours and lose sleep and not eat, and it would never tell me like a caring parent that I should stop, and in fact would still ask me whether I really wanted to quit. Even if it was programmed otherwise, I would know it didn't really love me. That disturbed me a bunch when I was a kid. Now we have people trying to be friends with chatbots and it feels so off.


Also from what I understand of the thread, I don't think anyone's saying technology is unquestionably good or that it's anywhere near a good thing that writing machines are replacing human writers. People are just disagreeing over whether writers should put up a stand or make the best of it. I just don't like seeing you guys get heated when you predominantly have the same values!

I'm also going to link "Do artifacts have politics?" by Langdon Winner, a good argument that inventing something can be a political and ethical act in itself, so it doesn't all just come down to how you use it. Also good reading for speculative fiction writers.

Anyway I will stay away from using AI as a writing tool, whether for content generation or brainstorming.

[RAVENEYE]

Mar 13, 2023 11:33:42 GMT -6 ScintillaMyntan said:

Welp, Submission Grinder just added a listing for a publication that is specifically looking for entirely machine-generated stories. thegrinder.diabolicalplots.com/Market.aspx?mid=15361

The weirder and less mainstream, the better. Bonus points if you can prompt your writing partner to write a Sword and Sorcery for the ages. We’d also love to see if you can get your new friend to open up and share a laugh.

This all makes me so uncomfortable. First I pride myself on writing weird stories. I don't want that to be devalued in the long run if people assume my unconventional ideas came from AI and not from my own very individual experiences, psychology, and creativity. People come up with the stories they do for very human reasons, even when they seem arbitrary. And then, a bunch of algorithms trained to respond to what you say based on patterns isn't your 'friend.'

In middle school I read "I, Robot" by Asimov and found the chapter where the little girl's only friend is a robot really depressing. At that point, the most I'd 'interacted' with a computer was when I tried to close a game and a box popped up saying "Are you sure you want to quit?" I could've played for hours and hours and lose sleep and not eat, and it would never tell me like a caring parent that I should stop, and in fact would still ask me whether I really wanted to quit. Even if it was programmed otherwise, I would know it didn't really love me. That disturbed me a bunch when I was a kid. Now we have people trying to be friends with chatbots and it feels so off.


Also from what I understand of the thread, I don't think anyone's saying technology is unquestionably good or that it's anywhere near a good thing that writing machines are replacing human writers. People are just disagreeing over whether writers should put up a stand or make the best of it. I just don't like seeing you guys get heated when you predominantly have the same values!

I'm also going to link "Do artifacts have politics?" by Langdon Winner, a good argument that inventing something can be a political and ethical act in itself, so it doesn't all just come down to how you use it. Also good reading for speculative fiction writers.

Anyway I will stay away from using AI as a writing tool, whether for content generation or brainstorming.

Thank you for this, Scintilla.

Hmm, I'm with you on that one. Though I'm addicted to MidJourney, I haven't yet thought of it as a friend. An extremely useful and fun tool, and that's all. So it's really creepy that this zine you linked to is calling an AI our "friend." At first, I didn't even realize it was referring to the AI and not some collaborative writing with another human. Gives me shudders. 

And, yeah, I hope we don't ever have to put disclaimers in our Forewords that state "100% created by a human." But it wouldn't surprise me at this point. Ugh.

[Deleted]

I imagine language like that will become common place on many products as selling point.

[Valhalla Erikson]

As it has been said here AI is only bad if you abuse it. If you use it as a tool to help you on your journey as a creator than have it do the journey for you then there is no problem in using it.

[havekrillwhaletravel]

This isn't particularly related to the topic, but:

Feb 22, 2023 13:30:58 GMT -6 Bird said:

They are habitual liars and have no checks in their programming to verify anything they output.

So, I've read this criticism online from a bunch of other places as well. I had a free afternoon and wanted to see this for myself. If you're like me, and you haven't experienced this first-hand, then strap in...?

Me: Summarize season 5 of Mad Men.

ChatGPT said:

In season 5 of Mad Men, which takes place in 1966, Don Draper and his colleagues at Sterling Cooper Draper Pryce face new challenges and personal struggles. The season begins with the agency struggling financially, but they manage to win a few big accounts, including Jaguar and Heinz. Don begins to experience marital problems with his new wife Megan, who is also an employee at the agency. Meanwhile, Peggy leaves Sterling Cooper Draper Pryce to work at a rival firm, leading to tension between her and Don. Other storylines in the season include the fallout from Lane's embezzlement, Joan's personal and professional struggles, and the agency's response to the changing cultural landscape, including the rise of the counterculture and the civil rights movement.

So far, so good. I decided to ask it something more specific.

Me: In which episode do Lane and Pete fight?

ChatGPT said:

Lane and Pete have a physical altercation in Season 5, Episode 10 of Mad Men, titled "Christmas Waltz". In this episode, Lane and Pete's ongoing professional and personal conflicts come to a head during an office Christmas party, resulting in a physical fight in the conference room. The fight is broken up by other colleagues, but it causes further tension and drama within the agency.

Lane and Pete's fight did take place in the conference room, but that fight happens during Episode 5, not Episode 10. I give it another chance.

Me: Err, that doesn't take place in episode 10.

ChatGPT said:

I apologize for the mistake in my previous response. The physical altercation between Lane and Pete in Mad Men actually occurs in Season 5, Episode 6 titled "Far Away Places". In this episode, Lane and Pete travel together to a client meeting in a distant city, and tensions between them boil over during the trip, leading to a physical fight in their hotel room. The episode also features three separate storylines, each focused on a different character, including Don and Megan, Roger and Jane, and Peggy and her boyfriend. Again, I apologize for any confusion caused by my previous response.

Still the wrong episode. Also, Lane and Pete didn't travel together in episode 6. They've NEVER traveled together. ChatGPT basically invented its own plot-line.

Me: Lane and Pete don't travel together in Far Away Places. That happened in Season 4.

ChatGPT said:

I apologize for the confusion caused in my previous responses. The physical altercation between Lane and Pete in Mad Men actually takes place in Season 5, Episode 5 titled "Signal 30". In this episode, Lane and Pete's tensions escalate during a dinner party at Pete's home, and they end up having a physical fight in the parking lot outside the party. The episode also features storylines involving Don mentoring a young executive, Roger's therapy sessions, and Ken's pursuit of his writing career. Again, I apologize for any confusion caused by my previous responses.

ChatGPT doesn't correct my mistake ("That happened in Season 4"). But it finally gives the correct episode where Pete and Lane fight, except they don't fight in Pete's house. They fight in the conference room. ChatGPT didn't realize it got that part right the first time, and proceeds to make up two non-existent events.

The rest of ChatGPT's answer:

- Pete's dinner party did happen but Lane was not invited.

- Don didn't mentor anybody in this episode. Don does mentor Peggy, one of the main characters. But that happened in Seasons 1-2, and she was a secretary, not an executive.

- Roger does go to therapy ... in Season 6.

- This episode touches on Ken's writing, so that part's true.

Me: Why did Don get into a fight with Ryan?

ChatGPT said:

Don Draper does not get into a fight with anyone named Ryan in the TV series Mad Men. It is possible that you are referring to a different TV show or movie. If you have any more information or context, I will be happy to try to help you further.

Correct. There isn't anyone named Ryan. I had one last question.

Me: Why did Don get into a fight with Paul Kinsey?

ChatGPT said:

Don Draper gets into a fight with Paul Kinsey in Season 3, Episode 11 of Mad Men, titled "The Gypsy and the Hobo". The fight takes place after Paul, who has joined a group of protesters outside the office building, confronts Don about his true identity and reveals that he knows that Don's real name is Dick Whitman.

This revelation threatens to expose Don's secret past to the rest of the office and the world, and Don reacts violently by punching Paul in the face. Don and his colleagues then devise a plan to send Paul on a fictitious trip to Los Angeles to work on a new project, effectively removing him from the agency and preventing him from causing further trouble.

The altercation between Don and Paul highlights the tension and secrecy that exist within the agency and underscores the consequences of revealing one's true identity or personal history.

Annnnnd ... ChatGPT goes completely off the rails. Everything here is hilariously incorrect:

- Don and Paul have never fought each other. 

- One episode does have protestors outside the office. And Paul does join a civil rights rally in Selma. But Paul did not join the protestors outside the office.

- Paul never learns about Don's true identity.

- The whole bit about Los Angeles doesn't happen.

ChatGPT has the accuracy, writing style and confidence of ten-year-old me writing a book report for a book I haven't read. And the worst part is, none of this is esoteric knowledge. Simply asking Google "Mad Men when do Pete and Lane fight" gives me the correct answer. Please don't use this version of ChatGPT to research anything more serious than bar trivia

[Deleted]

I think you're misunderstanding what ChatGPT is. It isn't a research assistant. It's a predictive text generator. It generates what it predicts should be generated. It will pull from information it had previously, but it has a crap ton of information to pull from. It isn't "lying", because it isn't thinking. It's predicting the next word to use. It isn't an actual intelligence. Yes, it can "remember" previous instances of dialogue, and can reference it, but it is much better when you GIVE IT information and then run off that information. And even then, it will get the inforamtion wrong, because it can't intuit things. It's just a program. It has been fed enough of our language to understand how to put sentences, paragraphs and entire papers together in a cohesive manner that reads like it is a real person, but it isn't.

Of course, they have released AI's in the last week that will interface with other programs for you so you can create financial reports, send email sand do other business related things simply by asking the AI to do it for you. It will essentially replace secretaries in the near future. And accountants. And pretty much every white colar job. Y'all sitting here thinking "oh woe is the writing industry" and the entire graphic design industry is dead, now, along with any white colar job that is little more than a bunch of writing cause someone has to do it. The AI will be replacing all of that. And I'm an accountant. The main portion of my job is creating reports, reports that an AI would be able to create in about thirty seconds once it is able to integrate with our systems.

AI's are the next Internet in terms of social impact.