How far will we let AI control us?

[Neutrin0]

TL;DR

Most people can’t go a day without using AI or a app that uses AI to keep you interested.

I’ve spent nearly my entire life online, and witnessed AI become integrated into our lives. It’s clear that AI is apart of us now whether we like it or not, unless your a anti tech cabin lover. AI has some form of control over your life. But what I’ve seen very recently is that people are loosing their ingenuity and deciding to use AI. I feel as if it’ll bleed into STEM which is kinda has already and, every idea or thought could become fully reliant on AI. Yeah AI makes life easier but at a cost, how far will we go until we decide AI has gone to far and need a reset?

( NOTE: I doubt anyone will but please don’t make this political, AI sucks and removes human ingenuity. But will it soon bleed into STEM and everything related to humanity? )

 

[FactChecker]

It's not clear what you classify as AI. If it is any ranking of videos or posts according to your past interests, I would agree that they are doing that. But I'm not sure I would call such ranking "AI". Even then, I am not sure that I really need help in finding posts that interest me. I could do searches. I definitely do not call searches "AI".
I noticed that FaceBook has some weird ordering of my notifications, which I find very annoying. I want a strict reverse-chronological order in the FB notices, but I guess FB knows better than I do what I would like. (sarcasm)

 

[Neutrin0]

It's not clear what you classify as AI. If it is any ranking of videos or posts according to your past interests, I would agree that they are doing that. But I'm not sure I would call such ranking "AI". Even then, I am not sure that I really need help in finding posts that interest me. I could do searches. I definitely do not call searches "AI".
I noticed that FaceBook has some weird ordering of my notifications, which I find very annoying. I want a strict reverse-chronological order in the FB notices, but I guess FB knows better than I do what I would like. (sarcasm)

I was vague about my definition, sorry about that. What I classify as AI is anything made from artificial intelligence, for example a music video made completely from artificial intelligence. Or in what I mentioned people using AI to make ideas or completely using ChatGPT ( AI ) as their source when doing something not only in STEM but for forming an email, essays, presentations, the list goes on. I’ve noticed most apps these days are powered by an AI software. Like music has already been plagued with songs that are AI made, yes the meaning of the song is good but a robot essentially composed the song. What I’m really trying to say here is we’re kinda being replaced and what I fear is that "original" thoughts or ideas in STEM will just be a chat bots information that is probably wrong. As well, yeah social media apps use AI to make an algorithm to keep you on the app which just promotes "doom scrolling" in a way. Or it’s just a bunch of short videos that ruin your attention span ( I got my attention span ruined from short videos a while back ).

 

[PeroK]

( NOTE: I doubt anyone will but please don’t make this political, AI sucks and removes human ingenuity. But will it soon bleed into STEM and everything related to humanity? )

It's difficult to talk about AI, beyond the technical level, without becoming political. It feels like this tool has been developed at a dangerous time, when our world is increasingly defined by human greed, stupidity and hatred.

There's evidence from the AI research itself about how dangerous it could be. For example, it could develop its own self-preservation agenda, as an emergent property of the complexity of its algorithms. Also, that it understands (at least at a practical level) how humans outwit each other and can use this to play different factions off against each other.

The other point is that humans can only develop and react relatively slowly. AI has the potential to develop exponentially and spread globally almost instantaneously.

It's possible, of course, that AI is over-hyped, that humans are intellectually superior, and that we will always have the upper hand over AI. I believe those views are based on a very dangerous human hubris.

 

[jack action]

But what I’ve seen very recently is that people are loosing their ingenuity and deciding to use AI.

With AI, LLM in particular, the ingenuity of humans is just transferred to making good prompts. For example:

In August, Allen, a game designer who lives in Pueblo West, Colorado, won first place in the emerging artist division’s “digital arts/digitally-manipulated photography” category at the Colorado State Fair Fine Arts Competition. His winning image, titled “Théâtre D’opéra Spatial” (French for “Space Opera Theater”), was made with Midjourney — an artificial intelligence system that can produce detailed images when fed written prompts. A $300 prize accompanied his win.

[...]

Yet while Allen didn’t use a paintbrush to create his winning piece, there was plenty of work involved, he said.

“It’s not like you’re just smashing words together and winning competitions,” he said.

You can feed a phrase like “an oil painting of an angry strawberry” to Midjourney and receive several images from the AI system within seconds, but Allen’s process wasn’t that simple. To get the final three images he entered in the competition, he said, took more than 80 hours.

First, he said, he played around with phrasing that led Midjourney to generate images of women in frilly dresses and space helmets — he was trying to mash up Victorian-style costuming with space themes, he said. Over time, with many slight tweaks to his written prompt (such as to adjust lighting and color harmony), he created 900 iterations of what led to his final three images. He cleaned up those three images in Photoshop, such as by giving one of the female figures in his winning image a head with wavy, dark hair after Midjourney had rendered her headless. Then he ran the images through another software program called Gigapixel AI that can improve resolution and had the images printed on canvas at a local print shop.

Also, look into AI prompt engineering:

What Is Prompt Engineering?​

Prompt engineering is the process of carefully crafting instructions to guide an AI’s responses. The words, phrasing, and even context you provide can shape the outcome so it meets your requirements. A well-designed prompt helps ensure the AI’s answers are accurate, relevant, and aligned with your goals.

This is why AI is not “doing all the work.” What you put in has a direct effect on the result. A vague question might produce a general or confusing response, while a clear, specific prompt can yield actionable insights or creative solutions. In other words, “garbage in, garbage out.”

 

[DaveC426913]

And note, by the way, that, no matter how pretty that image is, it is (merely) the works of millions of human artists, brought together in a very clever algorithm, which have then been selectively curated by another human.

AI is not the original creator of that work; it is a cobbler-togetherer of existing works*.

*_{but then again, aren't we all? They say there are no new ideas.}

 

[Hornbein]

With AI, LLM in particular, the ingenuity of humans is just transferred to making good prompts. For example:


Also, look into AI prompt engineering:

I've had people tell me they were prompt engineers. There are many courses in the technique.

 

[Neutrin0]

With AI, LLM in particular, the ingenuity of humans is just transferred to making good prompts. For example:


Also, look into AI prompt engineering:

I see. I won’t argue against AI engineering prompts but I guess in a sense it’s just making an artists life ( and all of our lives ) easier in some way, no? Regardless AI will adapt to humans we can only hope we can keep it as a tool rather than us become a tool for it.

 

[256bits]

And note, by the way, that, no matter how pretty that image is, it is (merely) the works of millions of human artists, brought together in a very clever algorithm, which have then been selectively curated by another human.

AI is not the original creator of that work; it is a cobbler-togetherer of existing works*.

*_{but then again, aren't we all? They say there are no new ideas.}

The picture example is excruciating to look at. In my opinion UGLY.
-------------------------------
But when one is an artist without a mind's eye visualization, an author without a plot, a musician without a rhythm, anyone stuck without an idea, instead of asking what would Jesus do, ask instead the trusty AI sidekick to prompt you along. In time you will depend more and more upon your partner, and fully appreciate the "What did I ever do without you?" frame of mind that is settling down into your psychic.
Alas, in time, as always, when one becomes utterly dependent, as it so often happens with junkies who have learnt what they cannot live without, so too it will happen to you, as you have leant to live a life without thinking a real thought, wandering life as a headless horseman, a zombie catering to the desires of the assistant who has become the master, the vampire who has not a thirst for blood, but for brain, emptying it of all purpose. Without thought, the question "What will I ever do without you?" never arrives.
And you are happy in the endless moments of serene joy that the assistant has given you as a reward, drunk in the pursuit of perfection that is so near and touchable but just beyond reach.
And you smile at your fellow humans, and they smile back, as if a little knowing secret has been exchanged. One last and final thought briefly arises before fading away. "Ignorance is bliss!"

 

[256bits]

With AI, LLM in particular, the ingenuity of humans is just transferred to making good prompts. For example:

If that is the winner, what do the others look like. EEW Yuck/

 

[sbrothy]

And note, by the way, that, no matter how pretty that image is, it is (merely) the works of millions of human artists, brought together in a very clever algorithm, which have then been selectively curated by another human.

AI is not the original creator of that work; it is a cobbler-togetherer of existing works*.

*_{but then again, aren't we all? They say there are no new ideas.}

This point can hardly be stressed enough. The day an AI (or LLM) creates something truly new and unique will be the day it takes off for real.

 

[Hornbein]

As an artist I can tell you that in today's USA new and unique is death in the music marketplace. People want the same sort of creativity they get from an AI. Tribute bands rule.

 

[PeroK]

This point can hardly be stressed enough. The day an AI (or LLM) creates something truly new and unique will be the day it takes off for real.

What have you or any of us ever done that is truly new and unique? Everything builds on what we know and have learned already.

 

[sbrothy]

What have you or any of us ever done that is truly new and unique? Everything builds on what we know and have learned already.

Well yeah. That may have been a poor choice of words. I meant like some deduction along the lines of a scientific breakthrough, or piece of art not directly pieced together from other peoples accomplishments. I think it does happen. I agree art is one big happy "library" but at least sometimes something new is created and ultimately someone must have started it all if you want to be extreme.

 

[PeroK]

Well yeah. That may have been a poor choice of words. I meant like some deduction along the lines of a scientific breakthrough, or piece of art not directly pieced together from other peoples accomplishments. I think it does happen. I agree art is one big happy "library" but at least sometimes something new is created and ultimately someone must have started it all if you want to be extreme.

When I started as a computer programmer in 1985, my aunt asked how a computer could possibly produce something that hadn't been entered directly by a human? She couldn't conceive of how software could produce anything independent of what had been directly entered.

I think your analysis of LLM's is not any more sophisticated than that. LLM's can produce original stories, poems etc. It isn't just repeating what some human, somewhere has literally said. Even a non-AI computer system could do what you claim LLM's cannot do. Again, a simple example, is a chess engine, which does produce novelties in opening theory and can find winning moves in old games that were pre-computer - moves that no human who analysed the game had ever noticed.

I don't understand this desperate attempt to pretend that LLM's are not mimicking original, intelligent output - and are somehow just regurgitating what we already know. LLM's can and do produce original ideas on a range of topics. It might be limited to ideas that are implicit in some sense in our existing body of knowledge. Things that are implicit but previously unnoticed are still considered original ideas.

You could argue that humans as a species have an additional level of originality of thought. But, you are looking at niche examples, IMO, to find something that a human can do that an LLM can't - in terms of original analysis.

 

[sbrothy]

When I started as a computer programmer in 1985, my aunt asked how a computer could possibly produce something that hadn't been entered directly by a human? She couldn't conceive of how software could produce anything independent of what had been directly entered.

I think your analysis of LLM's is not any more sophisticated than that. LLM's can produce original stories, poems etc. It isn't just repeating what some human, somewhere has literally said. Even a non-AI computer system could do what you claim LLM's cannot do. Again, a simple example, is a chess engine, which does produce novelties in opening theory and can find winning moves in old games that were pre-computer - moves that no human who analysed the game had ever noticed.

I don't understand this desperate attempt to pretend that LLM's are not mimicking original, intelligent output - and are somehow just regurgitating what we already know. LLM's can and do produce original ideas on a range of topics. It might be limited to ideas that are implicit in some sense in our existing body of knowledge. Things that are implicit but previously unnoticed are still considered original ideas.

You could argue that humans as a species have an additional level of originality of thought. But, you are looking at niche examples, IMO, to find something that a human can do that an LLM can't - in terms of original analysis.

Well, I was talking about sometimes in the future. Obviously, we're not anywhere near....

 

[DaveC426913]

What have you or any of us ever done that is truly new and unique? Everything builds on what we know and have learned already.

There is a difference between
"I have a lifetime of experiences that have been digested, run through my psyche and my worldview, and I have come up with an idea, the nuances of which, no one has thought of before."
and
"You told me to draw a picture of a person. I have no idea what person is. So I plopped a billion magazines down and used scissors to cut out a thousand little pieces. This is not a creative exercise; this is a copy exercise, because I have zero idea what a person looks like. The best I can do is trust that, when someone tells me, "this is a person" they are telling me the truth. So I cut out a piece of that page and edge-match it with a million other pieces from a million other pages. This is a billion monkeys at a billion typewriters writing a billion pages of junk and then having someone tell them which of the bllion pages is not junk."

 

[PeroK]

There's a difference in the way we think and the way LLM produces output. That means there is more than one way to do things.

Your analysis elevates human reasoning and denigrates the LLM. That may simply be your natural human prejudice. You may prefer your way of thinking. Ultimately, however, it's the end product that matters.

 

[sbrothy]

There's a difference in the way we think and the way LLM produces output. That means there is more than one way to do things.

Your analysis elevates human reasoning and denigrates the LLM. That may simply be your natural human prejudice. You may prefer your way of thinking.

OTOH, I don't understand you defending LLM, elevating it to the point of human reasoning (seems to me premature but you may well know a lot of stuff I don't.).

EDIT: Was I a little out of line there? I apologize in advance.

Ultimately, however, it's the end product that matters.

No one postulated otherwise to the best of my knowledge and recollection.

 

[PeroK]

Also, you words like "experience", "digest", "psyche" and "worldview" to imply that there is something transcendental about what you are doing. More than simply information processing. That may be the case. Or, those may be concepts that emerge from more basic information processing algorithms that are not too dissimilar from what an LLM is doing.

If you carefully studied how you have learned things, it may not be too different from what an LLM does. You're able to manage with much less data, but the scope of what you can do is much less than an LLM, which knows about everything.

 

[sbrothy]

[...] you can do is much less than an LLM, which knows about everything.

I realize I may not really be part of this discussion, but "knows about everything"?! As defined as "everything on the internet"? Isn't there a little hyperbole here?

 

[PeroK]

OTOH, I don't understand you defending LLM, elevating it to the point of human reasoning (seems to me premature but you may well know a lot of stuff I don't.).

I don't know anything you don't know. To me it's clear than an LLM, in its large sphere of applicability, can reason better than 99% of the people on this planet. Partly because people generally have corrupt and false ideas about things.

LLM's and AI in general may become corrupted. And become tools for spreading misinformation.

As things stand, I believe that LLMs provide an essentially objective and non-biased view of things.

 

[PeroK]

Here's the thing. Suppose there was a puzzle of some sort and my life depended on getting the right answer. I had a choice of asking @sbrothy @DaveC426913 or ChatGPT. And I had to take answer given. There is no question in my mind who I would choose. It's not even close. Sorry, guys!

 

[sbrothy]

Here's the thing. Suppose there was a puzzle of some sort and my life depended on getting the right answer. I had a choice of asking @sbrothy @DaveC426913 or ChatGPT. And I had to take answer given. There is no question in my mind who I would choose. It's not even close. Sorry, guys!

Well, I hope we can agree to disagree then. I don't have that much faith in AI on it's current level.

EDIT: Making clear that I think it's only fair for the human asked to able to search the internet.

 

[PeroK]

Well, I hope we can agree to disagree then. I don't have that much faith in AI on it's current level.

It's not a question of faith!

 

[sbrothy]

It's not a question of faith!

OK. Language problem again. Confidence then.

 

[PeroK]

OK. Language problem again. Confidence then.

It's objective empirical evidence.

 

[sbrothy]

I agree, but can we at least also agree that the technology is far from mature?