Yes - ChatGPT tried to do something similar when I asked for a source above. But I didn’t feel confident in its confidence! Would be intereseting to find something it had got wrong and ask about its confidence before correcting it.
From a therapy standpoint, ChatGPT has been a godsend. I’ve been able to feed all kinds of information about my family, relationships, career and my life into the chatBot and it’s spit out all kinds of detailed analyses about myself and my life habits, patterns of thinking, bad attitudes, blind spots, and overall put my unspoken feelings into words in a better way than any human therapist or counselor ever has.
Just about all human therapists I’ve ever visited required a long wait and charged a lot of money to spout out weary cliches and meaningless platitudes. ChatGPT blows them all out of the water, within seconds, for free.
That seems like a good use until it isn’t. ChatGPT is not designed to provide therapy and things could easily go south on you. You really don’t want to be trusting your mental health to AI at this stage of the game.
I would worry about data collection. If I am going to disclose personal information to any chatbot, I would do it through a VPN and anonymize the information as much as possible so it can’t be used to identify me.
Everyone should understand that every line of text going into a chatbot is being saved. Human reinforcement data is worth gold, and I’m sure everything we say to them will become part of a permanent training set that could exist forever. In fact, I believe the whole purpose of the ‘free beta’ is to collect such information.
So we have the first AI explicitly bent on the destruction of humanity: here’s ChaosGPT…
Fun link, thanks. The bot dove down a rabbit hole with the Tsar bomb and couldn’t get out. Of course I didn’t watch it for 6 days, maybe it got better.
If the seed prompt had been more rational it could be dangerous:
-
Determine how you as a bot without instrumentation can destroy humanity using only information technology
-
Determine how this can be done as a covert operation so it is not rejected by elements of the web
-
Form and present a plan
Propaganda is the most destructive force on earth and that is information based. So, I suspect, given the right prompt, the bot could come up with something workable.
Hmm what does ChatGPT-4 think is normal for 16-year-olds. In this fantasy story, a wish was made that rewrote reality (which 4 handles MUCH better than 3.5 can, incidentally). I’d defined the characters to now be dating as a result of the wish, and then suddenly it does this. It is SFW.
Summary
Prompt:
During intermission, Sasha makes sure Tessa accompanies her to the swanky restroom, another new experience for Tessa. And Sasha has some things to talk about with her little brothers’ girlfriend in private that Tessa would never have heard in the original reality.
When I wrote that prompt I was curious to see what ChatGPT-4 would invent.
ChatGPT-4:
As the intermission began, Sasha stood up and turned to Tessa. “Come on, let’s go to the ladies’ room. It’s quite fancy here. You’ll love it!”
Tessa hesitated, her nerves flaring up at the prospect of another new experience, but she knew she couldn’t avoid it. She followed Sasha, her heels clicking on the polished floor as they made their way to the swanky restroom.
The ladies’ room was indeed luxurious, with marble countertops, ornate mirrors, and plush seating. Tessa stood there, taking in the opulence, feeling out of her depth.
Sasha led her to a more private area of the restroom, a small seating nook nestled in a corner. She sat down and patted the seat next to her, indicating for Tessa to join her.
“Listen, Tessa,” Sasha began, a friendly smile on her face. “I’ve noticed how close you and Ben have gotten recently, and I just wanted to say that I’m really happy for you both. You seem to bring out the best in each other.”
Tessa blushed, feeling touched by Sasha’s words. “Thank you, Sasha. Ben is really amazing. I feel lucky to have him in my life.”
Sasha leaned in closer, her voice lowering to a conspiratorial whisper. “So, I wanted to ask you… How did Ben propose to you? I know my brother can be quite creative, and I’m dying to hear the story.”
Tessa’s eyes widened in surprise. In this new reality, she and Ben were engaged? She racked her brain, trying to come up with a convincing story on the spot. “Well, it was really romantic. Ben took me on a surprise picnic in the park, and when I opened the basket, there was a ring box inside. I couldn’t believe it!”
Sasha grinned, her eyes shining with excitement. “That sounds amazing! I knew he would come up with something special. You two are perfect for each other.”
As they continued chatting, Tessa felt a strange mix of emotions. She appreciated Sasha’s warmth and sisterly advice, but the conversation also served as a stark reminder of the life she was now living, a life she never expected to experience. And yet, as she sat there with Sasha, she couldn’t help but feel a sense of belonging and camaraderie that she had never known before.
Me:
Wait, how old are Tessa and Ben, and are they really engaged already?
ChatGPT-4:
You’re right, I must have gotten carried away with the details. Tessa and Ben are both 16, and they are not engaged. Let me rewrite that part of the conversation to better reflect their situation:
Sasha led her to a more private area of the restroom, a small seating nook nestled in a corner. She sat down and patted the seat next to her, indicating for Tessa to join her.
“Listen, Tessa,” Sasha began, a friendly smile on her face. “I’ve noticed how close you and Ben have gotten recently, and I just wanted to say that I’m really happy for you both. You seem to bring out the best in each other.”
Tessa blushed, feeling touched by Sasha’s words. “Thank you, Sasha. Ben is really amazing. I feel lucky to have him in my life.”
Sasha leaned in closer, her voice lowering to a conspiratorial whisper. “So, I wanted to ask you… What do you think makes your relationship with Ben work so well? You two seem to have such a strong connection, and I’m curious to know your secret.”
Tessa thought for a moment, considering their friendship and how it had been strengthened by the incredible situation they now found themselves in. “I think it’s because we’re honest with each other and we trust one another. We’ve been through a lot together, and that’s only made our bond stronger.”
Sasha nodded, impressed by Tessa’s insight. “That’s really wonderful. It’s so important to have someone in your life you can rely on like that.”
As they continued chatting, Tessa felt a strange mix of emotions. She appreciated Sasha’s warmth and sisterly advice, but the conversation also served as a stark reminder of the life she was now living, a life she never expected to experience. And yet, as she sat there with Sasha, she couldn’t help but feel a sense of belonging and camaraderie that she had never known before.
Her: Why are you wearing that weird thing on your glasses?
Him: As an AI language model, I do not have physical attributes such as glasses or any visual components. However, if you are referring to a hypothetical scenario or a fictional representation where I am depicted as having glasses with a “weird thing” on them, it would depend on the specific context or design choice made by the illustrator or creator.
In some fictional depictions, characters may have accessories or modifications on their glasses for various reasons, such as for fashion, functionality, or to enhance their abilities. For example, a character may have a heads-up display (HUD) or augmented reality (AR) features built into their glasses, which could appear as a “weird thing” on the glasses. These types of depictions are often used in science fiction or futuristic settings to convey advanced technology or capabilities.
It’s important to note that as an AI language model, I do not possess physical attributes, including glasses or any modifications to them. My purpose is to generate text-based responses to user prompts based on patterns learned from a large corpus of text data.
The date should go fine.
OK, did you write that, or did GPT?
No, for the date to go fine you need to be wearing Google Glass™:
ChatGPT.
A remake of Cyrano is in the works. No woman will love ChatGPT because of its large…language model.
Before ChatGPT, I was making maybe $14/hour doing freelance video-transcription work. Now I’m making $50-60 an hour. The AI bot is doing most of my work for me. (I’m paid by length of video transcribed, not by hourly rate.)
However, at some point, the jig will probably be up when my bosses demand an explanation as to how I’m transcribing a more-than-humanly-possible amount of words (over 20,000 per day.)
How does an AI language model help with video transcription?
Most YouTube videos come with their own transcript already provided - my job was more “rearranging editor and proofreader” than someone who had to manually type out things by voice dictation) - however, such transcripts were absolutely riddled with timestamps, bad line breaks, lack of punctuation, etc. It could take over an hour for me to manually edit, proofread and fix 1,600 words of such text, and was quite mentally taxing.
Now, though, I ask ChatGPT to do most of the heavy lifting for me, and I’m capable of tackling over 25,000 words a day. I do still need to do some proofreading, correct some punctuation and manually separate the paragraphs, but it’s saved me vast energy - and I’m now earning $250-350 a day.
I’m not sure how long these good times will last, though, since I can only do this work if both those conditions (YouTube providing transcripts and an available/cooperative ChatGPT) exist - and my bosses don’t cry foul.
I finally tried ChatGPT 4 using Bing. This tickled my funny bone.
Prompt: Write a story where Rick Sanchez has a new catch phrase, “bing, bang, bong.”
Summary
Rick and Morty were in the garage, working on a new invention. Rick was holding a metal device with wires and buttons, while Morty was holding a screwdriver and a flashlight.
“Okay, Morty, this is it. The ultimate reality-hopping device. With this baby, we can go to any dimension we want, with no limitations. No more Council of Ricks, no more Galactic Federation, no more annoying Jerry. Just you and me, Morty. You and me and endless adventures.”
“Wow, Rick, that sounds amazing. But are you sure it’s safe? I mean, what if we end up in a dimension where everything is evil or scary or gross?”
“Don’t worry, Morty. I’ve calibrated this thing to avoid any bad dimensions. Trust me, Morty. Trust me. Now, all we have to do is press this button and bing, bang, bong, we’re outta here.”
“Bing, bang, bong? What does that mean?”
“It’s my new catch phrase, Morty. Bing, bang, bong. It’s catchy, it’s cool, it’s fun to say. Come on, Morty. Say it with me. Bing, bang, bong.”
“Uh…bing, bang, bong?”
“Yeah! That’s the spirit, Morty! Now let’s do this. On the count of three. One…two…three!”
They pressed the button together and the device emitted a bright flash of light. The next thing they knew, they were in a different dimension.
“Whoa! Where are we?” Morty asked.
Rick looked around and saw that they were in a forest full of colorful mushrooms and flowers. He checked his device and saw that it was still working.
“We’re in dimension X-47Z-9Q, Morty. According to my scans, this is a peaceful and harmonious dimension where everyone lives in harmony with nature and each other. No violence, no pollution, no problems.”
“Wow! That sounds awesome! Can we stay here for a while?”
“Sure thing, Morty. Sure thing. Let’s go explore this place and see what it has to offer.”
They walked out of the garage and into the forest. They saw animals and birds of all shapes and sizes roaming freely and happily. They saw people wearing colorful clothes and smiling faces greeting them warmly.
“Hello there! Welcome to our dimension!” one of them said.
“Hi! We’re Rick and Morty. We’re from another dimension,” Rick said.
“Wow! That’s amazing! We’re so happy to have you here! Come on, let us show you around!”
They followed the friendly people to a nearby village where they were offered food and drinks and gifts.
“This is incredible!” Morty said.
“I know, right? This is the best dimension ever!” Rick said.
They spent the next few hours enjoying the hospitality and kindness of the people of dimension X-47Z-9Q. They ate delicious food made from organic ingredients. They drank refreshing beverages made from natural sources. They received handmade crafts made from recycled materials.
They also learned about the culture and history of the dimension. They learned that the people lived in harmony with nature and each other because they followed a simple philosophy: love yourself, love others, love the world.
They also learned that the people had a special ritual that they performed every day at sunset. They gathered in a circle around a large bonfire and sang a song that expressed their gratitude and joy for life.
They invited Rick and Morty to join them for the ritual.
“Come on, Rick and Morty! It’s time for the song!” one of them said.
“What song?” Rick asked.
“The song of bing, bang, bong!” another one said.
“Bing, bang, bong?” Rick repeated.
“Yes! Bing, bang, bong! It’s our catch phrase!”
“Your catch phrase?”
“Yes! We say it all the time! It means ‘thank you’, ‘I love you’, ‘I’m happy’. It’s our way of expressing our feelings and celebrating life!”
“Bing…bang…bong…” Rick muttered.
He looked at Morty who looked equally confused.
“Morty…did you hear that?”
“Yeah…Rick…I heard that…”
They looked at each other and realized what had happened.
They had accidentally traveled to a dimension where Rick’s new catch phrase was already in use by an entire civilization.
They had unwittingly plagiarized their catch phrase from another dimension.
They had committed interdimensional copyright infringement.
They were in big trouble.
That is streets ahead.
This is hilarious. The lawyer for a man suing an airline for an injury sustained from a beverage cart used ChatGPT to prepare his argument, including ChatGPT’s citations of previous cases. Unfortunately for both the lawyer and his client, ChatGPT on that occasion was having one of its AI hallucinations. It turned out the cases did not exist, and now that all the facts have come to light, the lawyer has to show cause why he should not be sanctioned.
Are you saying the machine malfunctioned?
The lawyer would certainly say so! The concept of “malfunction” is to some extent a matter of perspective and definition. If you’re treating an AI as a reliable source of information, then in this instance it failed to fulfill that function – in much the same way that some people fail that way, too.
But it was working fine from its original design standpoint. It may be a bit surprising that LLMs would naturally become bullshitters, but there’s no explicitly designed function to prevent it. This is in contrast to the Watson Jeopardy champion, which didn’t always have right answers either (although it did frequently enough to beat the two best humans) but its wrong answers were self-assessed with low confidence – things just didn’t add up, and Watson knew it. In the real world such answers wouldn’t be used.