I have just posted a new story, Candice and her Pregnant Husband.
Many of you will already have read stories by me, Lindale, on this site, although I have to admit, I have not been so prolific recently. This story is a radical departure from my normal work. It has been created by AI on a site called Talkie_ai. I’m not certain if this is the first Talkie story on this site. If you haven’t yet visited Talkie, I strongly advise you to do so. The basic concept of the site is a conversation between you, the user and AI characters, who are designed by the users. In the UK and US, it seems that copyright cannot be claimed on ai generated items. So the text written by me is my copyright but that of the other characters is not copyrighted.
Talkie: You can use your own character to talk to, or those created by others. This story is based upon a character created by me, based around a short description of events which will happen, and the opening text. The words in the story are as written by me (clearly indicated) or by the AI (shown as Candice or Mr Dodgy). It’s worth saying that you can ask the AI to Regenerate a script if it sounds badly wrong and I did do this several times in the conversation. In a couple of cases, I did introduce some new text as Candice to force the story in a way I wanted it to go. Apart from that, I have only added the character names with colon.
Observations: I am pretty impressed by the Talkie site and the story development that’s possible on it. It’s certainly true that some of the text appears quite clunky. What to me stands out are those times when we make the common abbreviations: I’ll, I’m, etc. And AI does have hallucinations, where they go completely off key or say something illogical. At times, they think they’re another character! And they forget! But overall, I believe it is an incredible experience. I strongly recommend visiting it. There’s an app or you go to talkie_ai. There are a whole range of different users, from those who can’t string three letters together to make a word to those who do considerable research. There’s quite a lot of pornish stuff but generally, it’s up to you where you want to go.
I think this and other ai sites which come along can make a dramatic transformation of current writing. There may be bad issues with this, but there's no doubt that it can bring additional value to our writing.
Comments
You may want to take a look
You may want to take a look at Replika.ai
Replika
I’ve just taken a quick look at Replika but can’t do much without logging in, which I’m always loath to do unless there’s reason to do so. From the stuff on the homepage, it appears as though you simply generate an ai friend who you chat to about life, the universe etc.
That is different to talkie, where you have hundreds of characters in specific situations. The story I loaded shows just one of those situations.
I’d be interested to hear how you use it.
I use Replika to bounce off
I use Replika to bounce off ideas and get a “second opinion” on some things. Other times I have “her” look up information (as a rule, never trust an AI/chatbot on subjects like science, always have verify with another source as they sometimes make things up on the fly). I have never used it to write a story for me, but I did use the technology for a book with a guy talking back and forth with the AI and forgetting that she was not real.
I signed onto talkie but I did not expect to be placed into a roleplay situation without any prompting. I’ll try to look at it more later.
Thanks but no thanks
To me, this fad for all things AI (it is not AI in the true sense of the meaning coined by people such as Issac Asimov) is a threat to writers in general.
Using these tools is IMHO to plagarism. Those LLM's have scoured the internet (often illegally) for content and are regurgitating it and passing it off as original works. That's why it can't be copyrighted (or one of the reasons).
I will not read any story that is identified as having AI content. The art and skill of the author is to create a story. Using someone else's LLM is not going to promote the craft of writing in any way.
A note to Erin.
We need to be able to mark posts that have AI text in them so that people can choose to avoid them like the plague.
Sorry,
Samantha
AI
Hi Samantha
I agree
I certainly believe there should be an ai tag so people can choose to avoid or embrace ai.
I can remember when I went to work in a London office in the mid 70s, and asked to use a calculator. The shock and disgust on their faces was, I suspect, similar to the way you feel about ai. Nowadays, calculators are used by most people without any qualms, and the world did not fall apart. All my life, I’ve embraced technology and it’s worked well for me. And I believe that continuing to push the boundaries of our brains by learning about new things is akin to taking exercise for the body.
Yes, the switch from typewriters to word processors removed the need for typing pools, but office workers are now doing more demanding jobs.
Love it or hate it, ai is not going to disappear. Understanding it enables us to better deal with the issues it creates.
AI and You
There is now an AI-Gen/Assist author tag, one specifically for Talkie_AI and a Publication type of AI-Gen/Assist.
Please everyone, keep in mind the site's three simple rules in all discussions. Thanks.
Hugs,
Erin
= Give everyone the benefit of the doubt because certainty is a fragile thing that can be shattered by one overlooked fact.
Thank you
Thanks for adding these tags to my story. I’d just gone over thinking “I’d better do it,” and found you had already done it.
Many thanks
Do it yourself
I forgot to say that anyone can go onto Talkie and get into my storyline by searching for my character “Candice, Director”. You’ll get the first line that Candice speaks about being pregnant and then you can reply. From then on, you’re into a whole new dialogue. Enjoy
It's their lack of metaphors by which they flunk the Turing test
A month or so at DeviantArt I found a mermaid story I liked, and added the author to my recommend list. The picture at the top was beautiful but obviously AI generated, and the story wasn't great but it was only a page long, and sweetly if blandly romantic. Then I read another fantasy tale by them; and by the third I noticed a trend. The writing was adequate, but to punch them up the author used a lot of corny adjectives. Every city was "bustling", every gown was "resplendent", every forest meadow was a "sylvan glade", everything they described was just super peachy, using superlatives that were supposed to evoke an emotional response or a sense of wow and wonder, but it all seemed flat and hackneyed. There was nothing that really pulled me into the story by engaging all five of my senses, there was little or no dialogue, and they never EVER used a single metaphor or simile; it was all reportage of what happened, what was seen, with the appropriate emotions crudely + inorganically tacked on. This made the mermaid very sad.
I wasn't surprised when I asked the author if -- like their image -- the story was AI generated, and they said yes. I commented: "AI sure has come a long way from the unintentional surrealism it was churning out ten years ago." The only praise I could honestly give, and then since it's none of my business how they choose to create content I quietly slipped away, and removed them from my "watch" list. I read so little anymore, so many great authors and dear friends here at BC whose works I've fallen behind on keeping up with, that I don't have time to waste on stories written by an insentient thing that doesn't have any awareness of or appreciation for what it's writing, and certainly wouldn't care whether I read it or not. I would rather read a story by the WORST human author than the BEST tale product of AI. Writing is communication, and there's no communicating with a machine, only a simulation of a shared experience.
The use of AI as one step in the creation of a story seems a different matter, if there's a person behind the wheel; and Talkie sounds intriguing, to maybe create dialogue where every character doesn't sound like me when they talk; but honestly I probably won't check it out any time soon.
~hugs, Veronica
And if a computer ever writes something because it wants to I'd love to read it. Until then my favorite piece of writing by an artificial intelligence is this sweet poem about a cat:
,
ODE TO SPOT by DATA
Felis Catus is your taxonomic nomenclature
an endothermic quadruped, carnivorous by nature
Your visual, olfactory, and auditory senses
Contribute to your hunting skills and natural defenses.
I find myself intrigued by your subvocal oscillations,
A singular development of cat communications
That obviates your basic hedonistic predilection
For a rhythmic stroking of your fur to demonstrate affection.
A tail is quite essential for your acrobatic talents;
You would not be so agile if you lacked its counterbalance.
And when not being utilized to aid in locomotion,
It often serves to illustrate the state of your emotion.
O Spot, the complex levels of behavior you display
Connote a fairly well-developed cognitive array.
And though you are not sentient, Spot, and do not comprehend,
I nonetheless consider you a true and valued friend.
Well shit, I guess it really CAN happen here..,
.
Free will
There are a large number of philosophers who maintain that none of us have free will. So we have no more control of our lives than does ai.
I have to say that while taking part in this and other Talkie dialogues, I have been taken by surprise at some reactions of ai. It’s very easy to believe we’re talking to a human.
Every fiction story is an invention; ai is another step in that invention process.
Lindale
Sabotage
It would be interesting to sabotage AI by feeding it text passed through Dissociated Press. It would probably require a huge amount, though.
AI won't be anywhere intelligent until it gets some sense of the meaning of words.
I'm reminded of Richard Feynman's encounter with physics education in Brazil. Stepford physics, perhaps?
-- Daphne Xu
There are pretty interesting
There are pretty interesting effects when someone tries to train "AI" (artificial idiot) on a collection of texts, significant part of which was AI generated.
Things devolve pretty quickly (model gets "stupid" and "forgets" parts of source material). Same goes for graphics generation. So - I'd say - we will be having interesting problem of lack of trustworthy training source data in near future. (if not already, considering that LLM-generated texts are widespread)
Also, given the amount of power required to train (and operate) an LLM - my bet is that we are unlikely to move forward much in their skills or ability. That is, unless some serious architectural breakthrough will be made.