#write with transformer

This project is unfinished and will remain that way. There are bugs. Not all endings are implemented. The ending tracker doesn't work. Images are broken. Nothing will be fixed. There's still quite a bit of content, though, so I am releasing what's here as is.

Tilted Sands is a project I started back when AI Dungeon first came out--the very early version you had to run in a Google colabs notebook. Sometime in late 2018, I think? I was a contributor at Botnik Studios at the time and I was delighted by AI Dungeon, but I knew it would never be a truly satisfying choose your own adventure generator on its own. I would argue that the modern AI Dungeon 2 and NovelAI don't fully function as such even now. That's not how AI works. It has to be guided heavily, the product has to be sculpted by human hands.

Anyway, it inspired me to use Transformer--a GPT2 predictive text writing tool--to craft a more coherent and polished but still silly and definitely AI-flavored CYOA experience. It was an ambitious project, but I was experienced with writing what I like to call "cyborg" pieces--meaning the finished product is, in a way, made by both an AI/algorithm/other bot AND a human writer. Something strange and wonderful that could not have been made by the bot alone, nor by the human writer alone. Algorithms can surprise us and trigger our creative human minds to move in directions we never would've thought to go in otherwise. To me, that's what actual AI art is: a human engaging in a creative activity like writing in a way that also includes utilizing an algorithm of some sort. The results are always fascinating, strangely insightful, and sometimes beautiful.

I worked on Tilted Sands off-and-on for a couple years, and then the entire AI landscape changed practically overnight with DALL-E and ChatGPT. And I soon realized that I cannot continue working on this project. Mainstream, corporate AI is disgustingly unethical and I don't want the predictive text writing I used to enjoy so much to be associated with "AI art". It's not. Before DALL-E and ChatGPT, there were artists and writers who made art by utilizing algorithms, neural networks, etc. Some things were perhaps in an ethical or legal grey area, but people actually did care about that. I remember discussing "would it be ethical to scrape [x]?" with other writers, and sharing databases of things like commercial advertising scripts and public domain content. I liked using mismatched databases to write things, like a corpus of tech product reviews that I used to write a song. The line between transformative art and fair use vs theft was constantly on all of our minds, because we were artists ourselves.

All of the artists and writers I knew in those days who made "cyborg art" have stopped by now. Including me.

But I poured a lot of love and thought and energy into this silly little project, and the thought of leaving it to rot on my hard drive hurt too much. It's not done, but there's a lot there--over 14,000 words, multiple endings and game over scenarios. I had so much fun with it and I wanted to complete it, but I can't. I don't want it to be associated in any way with the current "AI art" scene. It's not.

Please consider this my love letter to what technology-augmented art used to be, and what AI art could have been.

I know I'm not the only one mourning this brief but intense period from about 2014-2019 in which human creativity and developing AI technology combined organically to create an array of beautiful, stupid, silly, terrible, wonderful works of art. If you're also feeling sad and nostalgic about it, I hope you find this silly game enjoyable even in its unfinished state.

In conclusion:

1. You can live without cigarettes

2. It doesn't matter who you are

3. When there's a storm, a car chase, or war, remember we are the best at playing cards

4. You should go get your dog because he can tell your life story

5. I don't give two crap if your baby falls off the curb.

6. Be nice to the people who live next door and don't come to your funeral

7. You will be rewarded for everything you've done

8. I'm sick and tired of this shit

9. Do not get upset. Please stop the nonsense and go get some popcorn.

1. Priming

This is true across all predictive text models. The model needs a starting point, and can work with remarkably little (sometimes even a single sentence will do it!). It should establish the content, tone, and voice you want the model to generate further text in. On one of my earliest runs, I primed it with "I cast Slutify on you." I then proceeded to generate several lines of monologue describing the effects of this, in second person, gradually escalating as they went on.

Experiment! You may have to give it more to get at what you're after. Don't be afraid to just write more of your own content if it seems to be going in circles or need more direction. This roughly equates with responsiveness; giving feedback to your trance partner so they know what's working, what's not, and what to focus more on.

2. Patience and Choice The model is far from perfect, and will not always give you what you want it to right away. Prompting the model for completions will typically give you three choices. You may recognise this as the illusion of choice (do you want x or y?). Someone presents multiple options, which prompts a simple decision, but that someone is choosing to present specifically *those* options, and is discarding every other possibility. When we're working with someone we trust to make those decisions, it's a valuable tool for achieving clarity! It's also used by Dark UX design, but that's too much a tangent for this post.

What is relevant here is that this is not a real person, it's still just a model, and it's doesn't have the wherewithal to really know what choices to offer. This is another point where you're able to exercise responsiveness; if the model doesn't give you the options you want, just generate another set of responses. Depending on the configuration you can sometimes get bogged down in samey suggestions, but often you can get a pretty good variation.

3. Hypnotic Language

If you're after a trance-y experience, it helps to know the language keywords that trigger imagination to engage, and prompt the model with them. "Imagine," "what would it feel like if," "it's almost like," all give the model room to go off in new directions, and will help prompt the human mind to go off into speculative trance.

4. Dead-ending and Pruning

If things aren't working, be willing to undo. Sometimes you'll get a few paragraphs in and realize it's just not working and you need to prune it back a bit. Don't worry about it, this just happens. Remember it's just an experience, and that dropping a thread and returning to the conversational cul-de-sac is something we as humans do every day, and it's OK, it doesn't diminish the experience of the tangent.

5. Limitations of the model

As your text size increases you may notice the generative text gets shorter. This is because it takes more processing time the longer your primer text gets. Increasing the Max Time parameter will help to a certain size. After that, you just need to decrease your input size.

[NSFW]

Are you ready to finally get a man to come over to your table? Is your table getting lonely? Do you find your sex life lacking? You're not alone. Check out these reader-submitted stories and see if you can 't come up with one sexual move to impress your date.

From reader Janet, on what she calls "the most embarrassing time": "My man never stops complaining that his penis does a terrible job of penetrating. So instead, we watch one minute of the TV series Top Chef while he's trying his best. That's all! My advice is to remember that Top Chef is more than enough meat in any sandwich."

From reader Mary Ann: "Don't put your hands on your genitals too many times. Instead, put your hands on top of HIS genitals."

From reader John: "My boyfriend is constantly saying I look like a sexy nurse, so I started calling my dick the nurse's dick, and that gets some really nice reactions in his apartment when I do something really sexy with it. My advice is to always have an interesting gimmick for your cock and you will attract his attention."

From reader Katie on how to turn a guy's desire for fried chicken into a sexual encounter: "Fried chicken can be a great way to spice things up when you have sex! Start by spreading your legs to show him that your underwear is full of chicken, then use your thumb and forefinger to apply grease to his penis, nipples, or other areas where your partner likes to be smeared with flavor. He'll enjoy the opportunity to eat fried chicken, but be VERY careful not to let the food burn his nipples!"

From reader Michael: "A nice tip: hold his dick up high like a crucifix and  let him know you're a Christian, a good man, and you want an erection. Men find this really hot because all men worship Christ with all their heart!"

From reader Robert on the art of anal sex: "Most men don't know that the anus is made of nerves instead of solid bone. This is why the anus is sometimes called 'the Adam's apple of the genitals.’"

Which sexy tip are you most excited to try? Tell your friends about it! Click on the buttons below and then leave! Don't forget to keep checking for links to great products from Amazon!

As much as I liked this one, I had to try another AI.

So the following screenshots are all the AI generated content for different prompts.

I guess I could write this.

The space before the period truly makes this one.

What?

And then it just kept repeating the same thing. This is worse than my personal emotional breakdowns.

No honestly I don’t know what to say

My favourite part of this one is that save for the last sentence, it didn’t give me a choice and just picked a single sentence every time.