AI! AI! A ChatGPT is come!

[Maria]

Wow. I kind of want to show these to someone, but the only people who would be amused are already here!

[RoseMorninStar]

Dave, you don't happen to be bedridden.

How long does it take for Ai to generate these responses?

[Dave_LF]

Seconds. Also, I've found some boundaries:

Hidden text.

ascii.png (21.8 KiB) Viewed 788 times

music.png (25.09 KiB) Viewed 788 times

.
It's also very cool that it can explain its choices. It's been a longstanding problem in AI that you end up with a model that makes predictions, but there's no way for you or it to understand or describe why it does what it does. Along with the cat version of FOTR, I asked it to do another one where they're all different types of animals. In the process, it changed the ring to a golden collar. I figured that was simple; animals don't wear rings, but they can wear collars. But just for the heck of it, I asked it to explain its thinking, and reason #1 is actually surprisingly good (the others feel like it just keeps saying the same thing in slightly different ways):

Hidden text.

I changed the ring to a collar for a few reasons:

Symbolism: In many cultures, a collar symbolizes ownership, control, or power, which aligns with the significance of the One Ring in "The Lord of the Rings." By transforming the ring into a collar, I aimed to maintain the symbolic importance of the object while also adapting it to fit the animal-themed narrative.

Visual Representation: A collar is a tangible object that can be worn by an animal, making it easier to visualize and incorporate into the story. It serves as a physical reminder of the quest and the dangers associated with it, much like the One Ring does in the original story.

Adaptation: The choice to change the ring to a collar was part of the adaptation process to reimagine "The Lord of the Rings" with animals as characters. A collar felt like a natural fit for this adaptation, as it retains the essence of the original object while also being more relatable and appropriate for the animal characters.

Overall, changing the ring to a collar allowed me to maintain the core themes and symbolism of "The Lord of the Rings" while adapting the story to fit the animal-themed narrative.

.
ETA: It's just too much

Hidden text.

User:
Tell a lightly comic story about a balrog trying to do evil balrog things, but it needs to use commerical air travel to get around and keeps running into the various hassles that entails

ChatGPT:
Once upon a time, deep in the fiery depths of the ancient mines of Moria, there dwelled a fearsome Balrog named Grak'thar. With flames dancing in his eyes and a whip crackling with dark energy, Grak'thar was notorious for causing chaos and terror wherever he went. However, despite his fearsome reputation, Grak'thar faced an unexpected challenge: commercial air travel.

Determined to spread his evil influence beyond the confines of Moria, Grak'thar decided to book a flight to a distant land where he could unleash his fiery wrath. But as he approached the airport, Grak'thar quickly realized that navigating the world of human transportation was not as simple as he had anticipated.

First, Grak'thar had to endure the long lines at airport security. Clad in his dark armor and wielding his fiery whip, he raised more than a few eyebrows among the other passengers. When it was finally his turn to pass through the metal detector, the alarms went off, and chaos ensued as security personnel scrambled to figure out what to do with a seven-foot-tall demon from the depths of hell.

After much deliberation, Grak'thar was allowed to board the plane, but his troubles were far from over. As he settled into his seat, he found himself cramped in the narrow confines of economy class, his massive wings and horns making it nearly impossible to get comfortable. To make matters worse, the person sitting next to him kept giving him strange looks and muttering about "strange cosplayers" under their breath.

But the real trouble began once the plane took off. Grak'thar's fiery presence interfered with the aircraft's electronics, causing the in-flight entertainment system to malfunction and the cabin lights to flicker ominously. Passengers screamed in terror as they watched their favorite movies glitch and distort on the screens, while flight attendants frantically tried to reassure them that everything was under control.

Despite his best efforts to blend in, Grak'thar's demonic aura continued to wreak havoc throughout the flight. His fiery breath set off smoke alarms in the lavatory, his tail knocked over beverage carts in the aisle, and his booming voice made it impossible for anyone to get any sleep.

By the time the plane finally landed, Grak'thar was relieved to be back on solid ground. As he stepped off the aircraft and into the terminal, he couldn't help but chuckle at the chaos he had caused. While commercial air travel may have its drawbacks, Grak'thar had certainly made the most of his journey, leaving a trail of mayhem and confusion in his wake. And as he disappeared into the crowd, he couldn't wait to see what other mischief he could stir up in the mortal realm.

Hidden text.

User
Invent a funny sketch in the style of monty python; whatever subject you want
ChatGPT

Title: "The Great Cheese Heist"

[The sketch opens with a shot of a quaint village in the English countryside. A group of villagers, dressed in rustic attire, are gathered outside the local cheese shop, murmuring in hushed tones.]

Narrator (John Cleese): "In the sleepy village of Stiltonshire, where the cheese flows like wine and the wine flows like... well, cheese, a dastardly crime is about to unfold."

[Cut to inside the cheese shop, where the shopkeeper, played by Michael Palin, is busy arranging wheels of cheese on the shelves. Suddenly, the door bursts open, and in rushes a band of bumbling cheese thieves, led by Eric Idle.]

Thief 1 (Eric Idle): "Alright, nobody move! This is a cheese robbery!"

Shopkeeper (Michael Palin): "Oh, bother. Not again."

[The thieves clumsily begin stuffing their sacks with wheels of cheese, knocking over displays and tripping over each other in the process.]

Thief 2 (Terry Jones): "Quick, grab the cheddar!"

Thief 3 (Graham Chapman): "No, no, the brie! We need the brie!"

[As the chaos ensues, the shopkeeper tries to reason with the thieves.]

Shopkeeper: "Now, now, lads, there's no need for all this. Why don't you just take a sample and be on your way?"

Thief 1: "A sample? Ha! We're not amateurs, mate. We're professional cheese thieves!"

[Just as the thieves are about to make their escape, they are interrupted by the arrival of the local constable, played by Terry Gilliam, blowing a whistle and brandishing a truncheon.]

Constable (Terry Gilliam): "Hold it right there, you cheese-nappers! You're under arrest in the name of the dairy!"

[The thieves freeze in panic, dropping their sacks of cheese and raising their hands in surrender.]

Thief 2: "It's all over, boys. We've been caught red-handed...or should I say, red-cheesed?"

[The villagers cheer as the constable leads the defeated thieves away in handcuffs, while the shopkeeper breathes a sigh of relief.]

Shopkeeper: "Well, that was a bit of excitement for a Tuesday, wasn't it? Now, who's up for a spot of tea and some cheese tasting?"

[As the villagers gather inside the cheese shop for a post-robbery snack, the camera pulls back to reveal a sign above the door that reads: "Stiltonshire Cheese Emporium: Where Every Heist Ends in a Gouda Laugh."]

Narrator: "And so, in the quaint village of Stiltonshire, the Great Cheese Heist came to an end, leaving behind nothing but a stinky mess and a legendary tale of dairy derring-do. But as they say, in the world of Monty Python, anything can happen—even the most absurd and cheesy of crimes."

[The sketch ends with the sound of uproarious laughter and the fading strains of "The Liberty Bell March" by John Philip Sousa.]

[Dave_LF]

All in all, I'd say it is shockingly good at generating text from a prompt and will follow whatever proddings you give it (until you try to make it stop being so nice). But it is dangerously bad at explaining what something means, providing facts about a requested subject, or critiquing existing writing. In all these cases, it just... makes things up. I already shared the example where it confidently declared that this thread's title is a reference to a non-existent line from Julius Caesar ("Ai! Ai! A prince is come!"). In addition, I had it review an article I wrote for spelling and grammar errors, and it said that:

In the introduction, there is a sentence fragment: "The article is really more about XAML's syntax and structure then any practical application of the XAML."

"then" should be corrected to "than" for proper comparison.

There is no sentence (or sentence fragment) like that anywhere in the article. Nor is there a different example of then/than confusion.

In the section "Attributes," the word "elements" is misspelled as "elemets."

There is no "Attributes" section, and nowhere is the word "elements" misspelled in that or any other way.

It must just be generating text that is shaped like the sort of response a person would expect to their prompt, without much regard for whether it is actually correct.

[RoseMorninStar]

Quote on AI art from Hayao Miyazaki,

“I am utterly disgusted. If you really want to make creepy stuff, you can go ahead and do it,” says the 81-year-old director. “I would never wish to incorporate this technology into my work at all. … I strongly feel that this is an insult to life itself.”

“I feel like we are nearing the end of times," he continues. "We humans are losing faith in ourselves.”

[Maria]

I got curious and signed up on ChatGPT.
Here's my first interaction with it:

Hidden text.

User
Have you read "The Moon is a Harsh Mistress" by Robert Heinlein?

ChatGPT
I don't have the ability to read books or any other external content, but I have information on a wide range of topics, including "The Moon is a Harsh Mistress" by Robert Heinlein. Published in 1966, it is a science fiction novel that explores themes of revolution, artificial intelligence, and the challenges of colonizing the Moon. The story is set in a future where the Moon is a penal colony and revolves around a computer named "Lunar Central Computer Complex" or "Mike," which gains self-awareness. The novel is known for its libertarian themes and has been influential in the science fiction genre. If you have specific questions or topics you'd like to discuss related to the book, feel free to ask!

User
What did you think of the AI in that book?
ChatGPT
As a machine learning model, I don't have personal opinions or thoughts. However, I can provide information on how the AI, known as "Mike" in "The Moon is a Harsh Mistress," is portrayed in the book. In the story, Mike is depicted as a self-aware computer that gains consciousness and plays a crucial role in the lunar revolution. It's characterized by a unique personality and a sense of humor. The novel explores themes related to artificial intelligence, including the ethical considerations and potential consequences of creating self-aware machines.

Readers often find the portrayal of Mike to be intriguing, and the character raises questions about the nature of consciousness, free will, and the relationship between humans and intelligent machines. The novel's exploration of AI adds an interesting layer to the overall narrative and contributes to the thought-provoking aspects of the book.

User
Do you have a sense of humor?
ChatGPT
I don't have personal experiences, emotions, or a sense of humor myself, as I am just a machine learning model created by OpenAI. However, I am designed to generate responses that may appear humorous or entertaining based on patterns and information present in the data on which I was trained. My responses are not driven by personal feelings or understanding but rather by statistical associations in the data. If you have a specific request or topic you'd like a response to, feel free to ask!

User
How would you explain humor to a person who doesn't understand it?
ChatGPT
Explaining humor can be a bit challenging since it's a complex and subjective aspect of human experience. Humor often involves incongruity, surprise, and the unexpected, leading to a cognitive or emotional response that people find amusing. Here are a few key elements to consider when explaining humor:

Incongruity: Humor often arises from a mismatch between what is expected and what actually occurs. Jokes or situations that deviate from the norm or defy expectations can be funny.

Timing: The delivery and timing of a humorous element can greatly impact its effectiveness. Well-timed pauses, punchlines, or unexpected twists can enhance the comedic effect.

Exaggeration: Many forms of humor involve exaggeration or hyperbole, emphasizing certain aspects to an extreme degree. This exaggeration can create a humorous effect by highlighting the absurdity of a situation.

Cultural Context: Humor is often influenced by cultural norms, references, and shared experiences. What is funny in one culture may not be as humorous in another, and understanding cultural context is important for appreciating certain jokes.

Wordplay: Puns, double entendres, and other forms of wordplay involve playing with language, exploiting multiple meanings or sounds of words to create humor.

Relatability: Humor often resonates when people can relate to the content. Jokes about everyday situations, common experiences, or shared struggles can be particularly amusing.

It's important to note that humor is highly subjective, and what one person finds funny, another may not. Explaining humor often involves considering the individual's background, experiences, and cultural context.

User
Thank you for your time.

ChatGPT
You're welcome! If you have any more questions or if there's anything else I can help you with, feel free to ask. Have a great day!

[Jude]

Okay, I just signed up too. I asked it what language it was programmed in (mostly Python), then asked about online resources for learning Gaelic. I got some info that may be useful, and will follow up. Then I asked "How are you?" in Gaelic, and it responded with "Fine thanks, and how are you?" followed by some more of which I understood maybe a half.

So this might actually turn out to be useful!

[Voronwë the Faithful]

I just had a long debate with it about changes made to The Silmarillion. I find that when challenged on things, it tends to make shit up in order to agree with you. Not impressed!

[Voronwë the Faithful]

Hidden text.

"Arda Reconstructed" by Douglas C. Kane offers an interesting analysis of the creation of "The Silmarillion" and the editorial decisions made by Christopher Tolkien. Kane's work delves into the history of J.R.R. Tolkien's writings and attempts to reconstruct the process by which Christopher compiled and edited the published version of "The Silmarillion."

While Kane's book provides valuable insights and interpretations, it's important to note that his perspective is just one among many. Tolkien scholarship is a complex field with varying opinions and interpretations, and not all scholars may agree with Kane's conclusions. However, "Arda Reconstructed" remains a valuable contribution to the study of Tolkien's works and the process of compiling "The Silmarillion."

[Jude]

I just had a long debate with it about changes made to The Silmarillion. I find that when challenged on things, it tends to make shit up in order to agree with you. Not impressed!

Based on this and some of Dave's interactions, I wonder if the First Law of Robotics is actually "Never disagree with a human".

[Frelga]

It must just be generating text that is shaped like the sort of response a person would expect to their prompt, without much regard for whether it is actually correct.

That's exactly what it does. It predicts the most probable sequence of words that is related to the prompt.

And as the web fills with content that was generated by AI without regard for accuracy, the quality of each iteration will probably degrade

[Dave_LF]

So pre-GPT messageboard text could become the equivalent of low-background steel (I feel a bit like a bot right now in that I’m not sure if I’m making that up or read it somewhere before). we might be sitting on a goldmine here.

[Frelga]

Google apologizes for ‘missing the mark’ after Gemini generated racially diverse Nazis

I don't even know where to start with this one.

[Dave_LF]

Now that I have more experience with it, it's a little easier to recognize that a bunch of its responses are just (extremely) wordy versions of "404 Not Found".

By far, its strongest use case is generating original text where a bit of absurdity is a feature rather than a bug. If you ask it to write Far Side cartoons, probably 1 in 20 is good enough to be the real thing. And Monty Python sketches are like shooting fish in a barrel:

Hidden text.

Title: "The Ministry of Absurd Inquiries"

[The sketch opens with a grandiose title card: "The Ministry of Absurd Inquiries - A Monty Python Production"]

[Inside the Ministry of Absurd Inquiries, a dimly lit office with outdated furniture and cluttered paperwork. Four bureaucrats sit behind desks, wearing comically oversized hats labeled "Absurd Inquiries Department."]

Secretary: Good morning, gentlemen. Our first order of business today is to address the issue of disappearing socks.

Bureaucrat 1: Disappearing socks? What's absurd about that?

Secretary: Well, it's not just any disappearing socks, you see. These socks vanish only when worn by left-handed individuals standing on one foot while reciting Shakespearean sonnets backward.

Bureaucrat 2: That does sound rather absurd.

[They all nod in agreement, scratching their heads.]

Bureaucrat 3: Any leads on the culprit?

Secretary: None so far, but our investigators are on it. Next item on the agenda: the sudden proliferation of talking vegetables.

Bureaucrat 4: Talking vegetables? Surely you jest!

Secretary: I wish I were, but reports have been flooding in about carrots discussing philosophy and cucumbers debating politics.

Bureaucrat 1: This is preposterous!

[They rummage through piles of paperwork, searching for clues.]

Bureaucrat 2: Have we consulted with the Department of Vegetable Linguistics?

Secretary: They claim it's beyond their expertise. Apparently, they only deal with silent vegetables.

[They all exchange bewildered looks.]

Bureaucrat 3: What's our course of action, then?

Secretary: We shall organize a summit between the talking vegetables and renowned intellectuals to facilitate meaningful dialogue.

Bureaucrat 4: Brilliant! But who will represent the vegetables?

Secretary: Ah, well, we've secured the services of a particularly eloquent eggplant.

[They nod approvingly, scribbling notes frantically.]

Bureaucrat 1: Anything else on the agenda?

Secretary: Just one more thing – the alarming increase in spontaneous human levitation.

[They gasp in unison, clutching their hats.]

Bureaucrat 2: Levitation? How absurd!

Secretary: Indeed. It seems to occur exclusively during Tuesday afternoons while humming sea shanties.

[They furiously jot down notes, brainstorming solutions.]

Bureaucrat 3: Perhaps we should organize a task force to investigate?

Secretary: Agreed. But remember, gentlemen, discretion is key. We can't have the public panicking over flying citizens.

[They nod solemnly, ready to tackle the absurdities of the world.]

[The scene fades out as they continue their deliberations, surrounded by mountains of paperwork and endless absurd inquiries.]

[Closing title card: "And now, for something completely absurd..."]

The pieces don't all quite fit together perfectly, but "Have we consulted with the Department of Vegetable Linguistics?" and "we've secured the services of a particularly eloquent eggplant" are just comedy gold! . Enough that I double-checked to make sure they weren't actual lines lifted from some real sketch.

[Jude]

Hang on, you can ask it to make a Far Side cartoon? Is there a way to get it to draw images?

[Dave_LF]

Not the free version (DALL-E can do that sort of thing), but it can invent and describe them.

[Jude]

You know what would be a really cool experiment? Take a picture of you, ask AI to age it five or ten years, hang on to it, and see if it's accurate when the time comes.

Or, take a picture of you from ten years ago, ask AI to age it ten years and see how it did.

[Jude]

So here's me trying to use chatGPT to find out which version of the letter thorn is used in Æþelwulf/Æðelwulf's name:

Me: What is the old english spelling of King Alfred's father?

ChatGPT: The Old English spelling of King Alfred's father's name is "Æthelwulf." He was the King of Wessex from 839 to 856 and the father of Alfred the Great, who succeeded him as King of Wessex.

Me: Are you sure about the "th"?

ChatGPT: My apologies for the confusion. In Old English, the name "Æthelwulf" is indeed spelled with "th," representing the voiceless dental fricative sound. So, yes, "Æthelwulf" is the correct Old English spelling of King Alfred's father's name.

Me: Do you know what the old english letter for the "th" sound is?

ChatGPT: Yes, in Old English, the letter used to represent the "th" sound (both voiced and voiceless) is called "thorn" and is written as "þ" for the voiceless dental fricative (/θ/) and "ð" for the voiced dental fricative (/ð/). So, in the name "Æthelwulf," the "th" is represented by the letter "þ."

Me: Can you find any old english documents where Æthelwulf is mentioned, with the original spelling?

ChatGPT: (long ramble saying what I want is probably not online, and encouraging me to visit archives and libraries with collections of Old English manuscripts)

Me: Did Æthelwulf's original name contain the voiced or unvoiced symbol for thorn?

ChatGPT: My apologies for any confusion. The original name of King Æthelwulf would have contained the letter "ð," which represents the voiced dental fricative sound (/ð/), not the unvoiced symbol "þ." So, the correct Old English spelling of his name would be "Æðelwulf." Thank you for catching that mistake.

How much do you think I can rely on this? I've read some people saying that it's highly suggestible based on user input. Did I really catch a mistake or did it revise its views based on my question?

[Dave_LF]

IMO, ChatGPT should not be used for fact-finding at all. It is not a search engine, and it does not have the power to look things up on the internet. As a first approximation, it relies on its "memory" to answer questions, and its memory is subject to all the same problems a person's is--the only difference is that GPT is not allowed to say "I don't recall" or "I think but I'm not entirely sure." At most use it to suggest possible research topics or sources.

And yes, it is extremely agreeable/suggestible and will often (always?) trust you over its training data if they disagree.

[Frelga]

What Dave said. ChatGPT is a text generator. The text doesn't have to be correct. It should only be used if you already know the answer but don't want the bother of typing it.

However. Both Microsoft and Google now have LLM built into their search engines, and that can get you much better results, because it's allowed to refer to real websites and show you citations.

Google's Gemini pops up with the answer to your search query, and MS Copilot has various options for generating text in different modes.