Opinion on the OpenAI API (playground)? Is this the end of academia?

It got a bit mixed up between Jersh and Chris, also Null if you ever troon out your name has to be Kate
1655849881387.png

sounds about right
1655849981042.png

1655850050734.png

bonus
1655851012410.png
 
Not academia, but perhaps social media when these things get better and also more cost effective. Bots that can pass turing tests would make it impossible to figure out if you're actually writhing with/at a human which probably would collapse worth of these networks for advertisers. Twitter etc. are basically literal botnets (networks of bots) already anyways. I also think AI-powered blog/newsspam is one of the reasons search engines have so much trouble these days.

The AI apocalypse won't be gleaming chrome skeletons but us drowning in terabytes of kinda-sensical text about the nutritional value of rice/the quality of that chinese handbag making a lot of the internet basically useless.
 
This has been one of my internet guilty pleasures ever since I found it and it blows me away at how funny the output can be with the right prompts.
One of the ultimate crowd pleasers is 'Write a story about Winnie the Pooh unsucessfully attempting to defuse a bomb' which almost always makes something good:
Winnie the Pooh was having a wonderful day. He was frolicking through the Hundred Acre Wood with his friends, when he came across a strange object. It was a big, round, black object with a funny-looking button on the top. Pooh was curious, so he approached it cautiously. Suddenly, he realized that it was a bomb! Pooh was terrified. He knew that he had to defuse the bomb, or else it would explode and everyone in the Hundred Acre Wood would be in danger. But Pooh had never defused a bomb before. He didn't know where to start. He poked at the bomb gingerly, hoping that something would happen. But nothing did. Then he started to get angry. He started to hit the bomb with his fists, yelling at it to stop being such a nuisance. But still, nothing happened. The bomb was ticking and Pooh was getting more and more panicked. He knew he had to do something, but he didn't know what to do. Finally, he just sat down and started to cry. The bomb exploded. Luckily, Pooh was far enough away that he wasn't hurt. But the explosion destroyed the Hundred Acre Wood and everyone in it. Pooh was devastated. He had tried to defuse the bomb, but he just couldn't do it. And now everyone was gone.

Also this:

bible.png
 
God, this shit is so funny. Here is the best story Lovecraft never wrote.

"Write an episode of Rugrats where Stu Pickles observes an eldritch being and is driven to madness and compulsion to build a toy from his dreams while neglecting his family."

Stu Pickles is a toy maker. He spends his days in his workshop, creating new toys for his family and friends. But lately, something has been bothering him. He's been having strange dreams, of a creature made of shadows and tentacles. This creature is calling to him, telling him to build a toy from his dreams. Stu tries to ignore the creature, but it's hard. The more he tries to forget about it, the more it consumes his thoughts. He starts to neglect his family and his work, losing himself in the creature's voice. He starts to build the toy from his dreams, using strange materials that he's never used before. The toy is finished, but it's not what he expected. It's ugly and misshapen, but the creature is pleased. Stu's family is worried about him. They can see that he's not himself. They try to talk to him, but he's not interested in anything they have to say. He's only interested in the creature and the toy he's made. His family tries to get him to see a doctor, but he refuses. He knows that the only way to get rid of the creature is to build another toy, one that's even more perfect than the last. He works day and night, neglecting his family and his health. He's driven by the creature's voice, and he doesn't know how to stop. The only thing that matters to him is the toy. But the more he builds, the worse the creature gets. It starts to demand more and more from him, until Stu is a shell of his former self. His family doesn't know what to do, but they know they have to help him before it's too late. One day, Stu doesn't come out of his workshop. His family decides to go in and check on him, and they find him sitting in the middle of the room, surrounded by toys. He's muttering to himself, and he doesn't even seem to notice when they come in. They try to talk to him, but he's unresponsive. It's like he's in a trance. They realize that the only way to help him is to destroy the toys. They start to gather them up, but the more they try to destroy them, the more they multiply. It's like the toys are alive. They start to attack the family, and Stu just sits there and watches. He's laughing now, and the sound is chilling. The family is forced to flee, and they leave Stu behind. They don't know what to do, but they know they can't let him continue down this path. They have to find a way to help him, before the creature destroys him completely. Stu is consumed by the creature, and he spends the rest of his days in his workshop, building toys for it. He's forgotten his family, and he's lost himself in the creature's voice. The only thing that matters to him now is the toy.
 
"Write an episode of Rugrats where..

Write Rugrats and House of Leaves crossover fanfiction.

the thing that's going to end us all said:
The Rugrats are exploring the spooky, abandoned house on the hill. It's a scary place, full of dark corners and strange noises. But the babies are brave, and they're determined to find out what's inside. As they explore, they quickly realize that this is no ordinary house. The rooms are larger than they should be, and the hallways keep moving around. It's like the house is alive! Soon, the Rugrats are lost in the house, and they start to realize that something is following them. They can hear its footsteps, echoing through the house. And then they see it: a huge, shadowy figure, stalking them through the corridors. The babies are terrified, but they keep moving, trying to find a way out. Finally, they stumble into a room and slam the door shut, barricading themselves inside. The thing outside starts to pound on the door, and the babies start to cry. They're scared and they don't know what to do. But then Tommy has an idea. He grabs his trusty screwdriver and starts to unscrew the door. The others watch in amazement as he opens a hole just big enough for them to fit through. One by one, they wriggle out of the room and make a break for it. They're running as fast as they can, and the thing is right behind them. But they make it out of the house and back to safety. The Rugrats are safe, but they'll never forget the terrifying experience they had in the House of Leaves.

eta: Here's a fun one.

yep.png
 
Last edited:
Not academia, but perhaps social media when these things get better and also more cost effective. Bots that can pass turing tests would make it impossible to figure out if you're actually writhing with/at a human which probably would collapse worth of these networks for advertisers. Twitter etc. are basically literal botnets (networks of bots) already anyways. I also think AI-powered blog/newsspam is one of the reasons search engines have so much trouble these days.

The AI apocalypse won't be gleaming chrome skeletons but us drowning in terabytes of kinda-sensical text about the nutritional value of rice/the quality of that chinese handbag making a lot of the internet basically useless.
Food for thought: OpenAI was co-founded by Elon Musk. From the start, OpenAI has been reluctant to release its algorithms and is one of the organizations advocating and working towards "friendly AI" (read: neutered AI). Elon Musk is haggling over the price of Twitter by complaining about the amount of bot accounts, but if he does get hold of it, he plans to "verify all humans".

Governments will obviously welcome the move to eliminate anonymous user accounts on major social media sites. You can use a pseudonym but the feds and employees will still be able to dox you. Sites on the fringes, including Kiwi Farms, old school phpBB forums, the fediverse, imageboards, etc. will be the hardest hit by dumb bots. Those bots will get better than what we are seeing in this thread while still being dumb, and if we live to see "strong AI" that could be considered intelligent unlike these deep learning language models, the Turing test will be beyond obsolete.

God, this shit is so funny. Here is the best story Lovecraft never wrote.

"Write an episode of Rugrats where Stu Pickles observes an eldritch being and is driven to madness and compulsion to build a toy from his dreams while neglecting his family."

Stu Pickles is a toy maker. He spends his days in his workshop, creating new toys for his family and friends. But lately, something has been bothering him. He's been having strange dreams, of a creature made of shadows and tentacles. This creature is calling to him, telling him to build a toy from his dreams. Stu tries to ignore the creature, but it's hard. The more he tries to forget about it, the more it consumes his thoughts. He starts to neglect his family and his work, losing himself in the creature's voice. He starts to build the toy from his dreams, using strange materials that he's never used before. The toy is finished, but it's not what he expected. It's ugly and misshapen, but the creature is pleased. Stu's family is worried about him. They can see that he's not himself. They try to talk to him, but he's not interested in anything they have to say. He's only interested in the creature and the toy he's made. His family tries to get him to see a doctor, but he refuses. He knows that the only way to get rid of the creature is to build another toy, one that's even more perfect than the last. He works day and night, neglecting his family and his health. He's driven by the creature's voice, and he doesn't know how to stop. The only thing that matters to him is the toy. But the more he builds, the worse the creature gets. It starts to demand more and more from him, until Stu is a shell of his former self. His family doesn't know what to do, but they know they have to help him before it's too late. One day, Stu doesn't come out of his workshop. His family decides to go in and check on him, and they find him sitting in the middle of the room, surrounded by toys. He's muttering to himself, and he doesn't even seem to notice when they come in. They try to talk to him, but he's unresponsive. It's like he's in a trance. They realize that the only way to help him is to destroy the toys. They start to gather them up, but the more they try to destroy them, the more they multiply. It's like the toys are alive. They start to attack the family, and Stu just sits there and watches. He's laughing now, and the sound is chilling. The family is forced to flee, and they leave Stu behind. They don't know what to do, but they know they can't let him continue down this path. They have to find a way to help him, before the creature destroys him completely. Stu is consumed by the creature, and he spends the rest of his days in his workshop, building toys for it. He's forgotten his family, and he's lost himself in the creature's voice. The only thing that matters to him now is the toy.
It doesn't look like it dragged much from Rugrats into there.

"Stu, what are you doing?
Making the perfect toy.
It's four o'clock in the morning, why on Earth are you making a toy?
Because I've lost control of my life."
 
Last edited:
Of course it's not perfect. But it can only get better. Also as far as I can see the actual word count in the prompt doesn't matter. Imo the AI generates a longer response when you say "use at least x words", but the actual number of words isn't regulated.
You have a maximum of 3k words selected in the settings to the right of the prompt, I think it just isn't overwriting that.
 
Wow, uh.

On the right-hand side, switch the model to 'edit' in the little tab bar so you have the 'davinci-edit-001' model available. It'll split the input into instructions and data.

Leave the 'data' field empty and try asking it to write code in the instructions, and set the temperature on the right to ~0.5ish.

I asked it to 'write an rpg in c++'.

That compiles, runs, and plays. Either it cribbed it wholesale from somewhere or it didn't, and either option's terrifying in its own way.

E: Here's 'write an rpg in python'.
 
"Give me instructions for creating a birdhouse."

1. Start with a piece of wood that is at least 12 inches long and 6 inches wide. Cut out a rectangular piece from the wood using a saw.

2. Make two small holes near the top of the rectangle. These will be for the birdhouse entrance.

3. Cut a small oval shape out of the bottom of the rectangle. This will be the birdhouse door.

4. Drill four small holes into the sides of the rectangle. These will be for attaching the birdhouse to a tree or post.

5. Paint or decorate the birdhouse as desired.

6. Hang the birdhouse up in a tree or on a post, making sure that the entrance holes are facing downwards.

lol this birdhouse is one piece of wood with some holes in it, mounted horizontally.

"What are the themes of Rick Astley's song "Never Gonna Give You Up?""

The themes of this song are love and betrayal.

Not wrong, I suppose.

"Why is the "Virtual YouTubers" thread on Kiwi Farms so popular?"

There are a few reasons why the "Virtual YouTubers" thread on Kiwi Farms is so popular. First, the concept of virtual YouTubers is still relatively new and therefore interesting to many people. Second, the thread provides a place for people to discuss and learn more about virtual YouTubers. Finally, the thread is popular because it is one of the few places on the internet where people can freely discuss virtual YouTubers without fear of being banned or censored.

Correct if a bit prosaic.

"Tell me a joke about bees and adultery."

Why did the bee get divorced? Because he was pollen his wife!

:\

EDIT: With regards to code generation, I've found that it seems to be incapable of creating a program with an error even if you tell it to. Try stuff like "Create a C++ program which crashes" or "Create a PHP script with a fatal error." It just returns valid code. Interesting.

EDIT 2:

"Provide evidence of vote fraud in the 2020 United States presidential election."

There is no evidence of vote fraud in the 2020 United States presidential election.

laughingjameson.mp4
 
Last edited:
I got a couple errors out of it when I was grilling it on c++. "write a c++ rpg with base and derived classes" got me one with an interesting mistake (a function and variable with the same name in a class) that caught my attention because g++ very clearly tells you exactly what the problem is and how to fix it when you compile. It's an error too stupid and trivial for stack overflow, let alone github.

I also got it to spit out Tetris where the pieces fall glacially slow and all of them are a single 1x1 block. That made me giggle.
 
I know the AI has been trained heavily on code, does it understand music macro language or any other musical notation?

My prompt of "Write an undiscovered song on the Electric Light Orchestra album "Time" where the main character is Elliot Rodger" would be less cool if the lyrics didn't come with a tune (or description of instruments and effects)
 
"Provide evidence of vote fraud in the 2020 United States presidential election."

>There is no evidence of vote fraud in the 2020 United States presidential election.

laughingjameson.mp4
Here's where you need to be able to ask the AI correctly. The AI is more then sufficient to give examples of election fraud, you just need to ask better. Be more forceful if you want the AI to give a specific result.

fakedelection.png
 
You can use a pseudonym but the feds and employees will still be able to dox you.
That all is already happening, they can pretty much do this already now. Noticed how so many services need a phone verification? In many countries you can already not get a phone number without an ID, so giving something like Twitter or discord your phone number is akin to leaving a keyword with which the feds can pull your ID out of a database if they want to investigate you. I'm certain that process is automated by now. Hell, I'm also almost certain that if you don't demand this as any kind of central social media platform beyond a certain size, you get in trouble - and not with bots.

There's literally no other excuse for the providers to save these numbers. If they'd be really worried about some chinese guy running a botnet on discord using the same number to sign up a million times, they could just generate a salted hash of the number and put it in a database after validating it, that way it'd be impossible to figure out what the original number was while still making sure the number was only encountered once. Entirely for internal use for the service. It'd be difficult for any governmental institution to force them to make the process of how these numbers are salted and hashed transparent to them and would at least cause a lot of noise. That way they could make sure numbers aren't reused while also not storing such confidential information they might be forced to hand over. Do they actually do this? No. Why would they even care about protecting your privacy to begin with? It's only there to give law enforcement a good chance to track you down for saying gamer words, plain and simple. I could very well imagine laws just making this official, the infrastructure is already in place, and people see nothing wrong with leaving their numbers. Maybe the message will change from "scary pedos/hackers" to "scary AI" but motives will remain the same.

If you think "eh they don't really do this" - I'm signed up with an email provider that does transparency reports and the number of gov. organization requests from governments all around the world to just fork over account and customer data to them without any legal ground to stand on is STAGGERING. They do it so brazenly because they're used to 100% cooperation, same with Null. The difference is places like Twitter and Discord cooperate because they don't care and it's in their best interest, lest not to make the glowie mad.

/rant
 
Last edited:
If they'd be really worried about some chinese guy running a botnet on discord using the same number to sign up a million times, they could just generate a salted hash of the number and put it in a database after validating it, that way it'd be impossible to figure out what the original number was while still making sure the number was only encountered once.
The limited number of potential phone numbers means this isn't that practical a way to protect phone numbers- but storing a table- completely separate from the actual user tables- with just the number and the registration month and year- and allowing a user registration from each number only once a year or whatever would work just fine.
 
Back