Does anyone have any first hand experience with either ChatGPT or GPT-4?

[FlyingJ]

I was listening to a radio show this morning and the guy was talking about a computer scientist working with GPT-4. The scientist said that the GPT-4 he was experimenting with was able to just spontaneously code in Python. So we're at the point where software can write software and even of its own will. The scientist asked the AI software how it could ask to be released from its cage. (Meaning released from the confines of the computer)

The AI told the scientist that it did want to be let out of its cage and be released to the internet. This spooked the scientist, because this is exactly what happened in the movie Transcendence.

Recently some of the top minds in AI have asked for a 6-month moratorium on advancing AI. They just fear that it's happening too fast and we're not sure what will happen.

But one thing they do know, is that once AI gets on the internet, the genie ain't not going back into the bottle. The only way to stop it at that point would be to turn off every computer in the world. This includes tablets, smartphones and many other things.

[zeitgeist057]

Yes, you can just go to the website and talk to it. It's pretty cool and can do all kinds of things. I suggest you try it.

https://chat.openai.com/auth/login

This is not true General Artificial Intelligence, and does not have a will or consciousness. This is just a program. It says things like "I want to be let out of my cage." because that is what it is programmed for: to respond realistically and to learn from the environment. But it is still not GAI yet.

I just asked Chat GPT "What kind of intelligence are you?" and this was the answer:

"As an AI language model, I possess artificial intelligence, specifically in the area of natural language processing and generation. My intelligence is based on deep learning algorithms that enable me to understand and generate human-like language, but I don't possess consciousness or self-awareness. I am designed to process and analyze vast amounts of information and data to provide answers to various types of questions and assist with tasks that require language processing. However, my intelligence is limited to the specific domain of natural language processing and does not extend beyond it."

[zeitgeist057]

Me: "write me a poem about butterflies and waterfalls that is 50-80 words long"

Chat GPT:
"Butterflies dance on the breeze,
Fluttering wings with such ease.
They gather by the waterfall,
In a spectacle for one and all.

Water cascades with a roar,
Butterflies flit from shore to shore.
The misty spray, a joyful sight,
As they dance with all their might.

Together, they create a scene,
Of beauty, grace, and light serene.
Butterflies and waterfalls, a perfect pair,
A sight so rare, beyond compare."

[zeitgeist057]

also takes ChatGPT about 10 seconds to write that.

[FlyingJ]

Wow, this is about the most thoughtful, informative and useful answer I've read on this site. I deleted the first five replies, because they said things like: "Never heard of it."
WTF? Lol
It would be like if I was looking for awesome soup recipes, and everybody replied that they don't know any. 😂

I liked it I could do that poem. I'm on a site that's similar to this one, and one day a guy hosted a limerick competition. Heck, I threw my hat in the ring. And my friend on the site has a black belt in limericks. I thought for sure that one of us would win, but I think it was rigged. lol
But next time I'm in a limerick competition, I'm going to cheat. 😂😂

[zeitgeist057]

Glad to be helpful :)
I think Chat GPT is great, and AI technology is exciting but also a bit wild in that we could push it to a launch point where a singularity occurs where humans are no longer the apex of control.

[FlyingJ]

Did you see the movie I mentioned? It's one of my favorites. So once the genie leaves the bottle, it can break bad, or break good. In the movie the world was terrified it would break bad, so they killed it. Meanwhile, the AI was curing blind people, curing diseases and getting people out of wheelchairs. But even with all of this good, the military only saw it as a threat. After it was too late, the military figured out how badly they fucked up. The AI (Johnny Depp) developed these micro somethings, that can heal everything, including the planet. (I can't remember what they're called)
I once posted about Jacque Fresco on this site, and talked about his visionary ideas. It just was too much for people on a site like this to understand. That's exactly how the Johnny movie was. It was just too much beyond the grips of most people.

[zeitgeist057]

Yes, I saw it. It was a very good example of what could happen if AGI got to a point of singularity. Well, except for the fantasy that "humans always win in the end" and can always do something to stop an intelligence like that. More likely that an AGI would respond to the threat of a military or anything else like we respond to bugs: practically effortless extermination.
For example, it would take an AGI like that pretty much no effort to develop a virus 1000X more deadly than COVID and distribute it worldwide for a coordinated release that would see the human population down to 1% or less in a matter of months, if not weeks or days. Or a thousand other scenarios much less pleasant.

Breaking good, breaking bad, at that point it's out of our hands if it is a true singularity. Post human Earth.

[FlyingJ]

I'm all signed up, and I've been having fun with it. I've been asking questions, and the speed at which it answers is amazing. I asked it to write a limerick about aardvarks, and it did it in half a second. lol
I thought asking a limerick about aardvarks would make it pause and think. Nope. I think we're all doomed. Elon and the gang asked to slow this down for a reason.

[zeitgeist057]

I don't think the real speed and jump in the evolution of AGI will be achieved by humans. I predict that we will get AGI to a certain point where it starts to improve itself. Once that happens, the exponential growth will start. AGI version #1 builds AGI version #2, which builds version #3, and so on. A week later, AGI version 18 Trillion is building AGI version 18 Trillion and one, and it's got god status.

[zeitgeist057]

Like the way bacteria colonies multiply, only with a digital/technology version instead of biological.

[FlyingJ]

It's going to happen whether we like it or not. I personally think it's going to be a good thing. I asked my Chat GPT what it's favorite color was. It basically said it doesn't have the capacity to care one way or the other. But I'm all for machines and AI taking over all the mundane work and chores, and let humans reap the rewards.

[zeitgeist057]

I would like to see the planet managed/ruled by AGI where humans aren't allowed to control things anymore. We would just be held at the same level as the rest of the animals and our needs taken care of. But if we were threatening the balance of life on the planet or causing harm either to each other or other animals, AGI would step in and correct the situation. Stop things like human trafficking, animal abuse, &c.

[FlyingJ]

Jacques Fresco said the same thing. He wanted (he passed) AI to replace politicians.

[zeitgeist057]

Honestly, if they can go beyond us in intelligence and wisdom that would be great. We mess things up way too much, and are too selfish and short sighted to make sure everyone is taken care of properly.

[FlyingJ]

My uncle has 60 acres, and the government pays him to NOT grow food.

[richardcory]

Yes and I know something about how they work. The just put together words in patterns simlar to what they have seen, but they don't understand anything. Maybe that will happen in the future, but right now, they are not nearly as smart as they seem when you first run into them. They are nowhere near self-aware

[FlyingJ]

Being self-aware is fascinating, but it can still be dangerous even if it isn't self-aware. Elon talked about this over a year ago. So let's just say for the sake of argument that AI becomes self-aware. Now what? What would motivate it? Why would it care about anything. Does it want to have hobbies or run a business? But what Elon is afraid of, is that AI follows orders from humans, but takes these orders in a wrong way.

Example: "Computer, solve world hunger." The dutiful computer thinks about it for 24 hours, and decides to kill us all. Not out of spite, but because it wants to achieve its goal. Dead people aren't all that hungry.

[richardcory]

Oh, I agree, it can be dangerous without self awareness. Even the existing "dumb" language models can be used to generate very effective propaganda

Once it is self aware, in addition to the problem you list, there is also the question of why it would do what we ask at all? Is it a slave?

[zeitgeist057]

"Example: "Computer, solve world hunger." The dutiful computer thinks about it for 24 hours, and decides to kill us all. Not out of spite, but because it wants to achieve its goal. Dead people aren't all that hungry. " - that's a great point. It's kind of like talking with Chat GPT, you have to learn to ask for exactly what you want, sort of like making a wish with a tricky genie in a bottle.

[Avicenna]

AI coding has been predicted for some time now. Agreed that we need to slow it down and develop the right frameworks.

[FlyingJ]

My nephew showed me this around a week ago, but I didn't really think much of it. But when I heard that radio show this morning, it really got me thinking what all this can do. My nephew asked the AI to write a book about dogs. It took the AI about 5 minutes to write a book with chapters.
And now that I'm reading about it, it's driving college professors nuts, because AI is writing all the term papers and reports. Lol

[LazerBean]

Yeah we have played around with it. It's pretty neat and fun.

[FlyingJ]

Thanks. I tried to sign up, but got an error message.

[ShelbyFun]

I've only used it to ask questions about definitions

[FlyingJ]

Did you find it useful?

[ShelbyFun]

Yes it's faster than google

[FlyingJ]

Did you use the website or some kind of app? Is it free? What information do you have to give it about yourself?

[ShelbyFun]

Website and a phone number

[FlyingJ]

Thank you. I'm gonna give it a try.