A.I. Photography - Are we doomed?

[snowbear]

This has been an interesting discussion. One important use of AI (and Machine Learning) is damage assessment from major storms. We (geospatial analysts and emengency managers) are able to assess damage and better determine where to place "boots on the ground" first.

Arguments have been cleared, and this is the only further warning to everyone: any more outbursts will cause the thread gets locked and time-outs issued.

 

[wobe]

I think pretty much everything is doomed with this fourth industrial AI revolution. Most people think it just means robots, they don't realise it will literally destroy everything, even human biology...

I actually wonder if AI and the great interconnection we have now is ‘The Great Filter’, one of the theories from the Fermi Paradox.

The world is unstable enough and getting much worse.
Now along comes an artificial imagination tool which aims to be so perfect as to become unrecognisable from reality (be it news, articles, photos or video to name but a few). The added chaos, misinformation, abuse and perversions stemming from this could so very easily lead to a very dark place societally.

 

[Original katomi]

Ok I have skipped through the past pages sky net and co aside:
as a hobbyist photographer and camera club member I used to take part in club level competition
so no I am starting to complete against A.I. images or A.I. enhanced images and not the skill of another person
in the future are we going to have human vs non human category’s for comp’s

 

[Hardus Nameous]

People still use horses, straight razors, campfires and paint. I still enjoy my cameras.

 

[ClintLloyd]

I actually wonder if AI and the great interconnection we have now is ‘The Great Filter’, one of the theories from the Fermi Paradox.

The world is unstable enough and getting much worse.
Now along comes an artificial imagination tool which aims to be so perfect as to become unrecognisable from reality (be it news, articles, photos or video to name but a few). The added chaos, misinformation, abuse and perversions stemming from this could so very easily lead to a very dark place societally.

Of course. The AI has been programmed by the very organisations that are orchestrating the whole stage show. It's genius, everyone will believe that the AI is all knowing, when it's actually all reciting...

Type in on your web browser 'the fourth industrial revolution' and you'll get plenty of information about the current situation.

 

[smoke665]

Ok I have skipped through the past pages sky net and co aside:
as a hobbyist photographer and camera club member I used to take part in club level competition
so no I am starting to complete against A.I. images or A.I. enhanced images and not the skill of another person
in the future are we going to have human vs non human category’s for comp’s

Seems that many clubs and competitive events have been proactive in their rules on AI images. There's been several incidents recently where images declared as winners have been disqualified, because of any AI editing including generative fill.

It's genius, everyone will believe that the AI is all knowing, when it's actually all reciting...

That's coming out now, look at the Google mess with their prejudicial programming. Sadly the code is not sterile, it's generated by humans, who conscious or not have an influence on the outcome. IMO even if, "God Forbid", the code becomes self generative, who's to say it won't develop it's own prejudices.

@JoeW you raise some very real concerns. AI by itself can be benign tool, which unfortunately can become a dangerous thing in man's hand. Because of that it will ultimately require regulation. In the digital world I would think it fairly easy to tag an image, story, etc with some type of registration mark tied to an authentication.

 

[wobe]

@JoeW you raise some very real concerns. AI by itself can be benign tool, which unfortunately can become a dangerous thing in man's hand. Because of that it will ultimately require regulation. In the digital world I would think it fairly easy to tag an image, story, etc with some type of registration mark tied to an authentication.

You talking about regulation in the US/West I presume? You think this regulation will be followed by every country?
Unless it is (which is extremely unlikely - especially as AI know how propagates) then the produce of AI will still be on the web to fool, mislead and sow discord to everyone, no matter where they are - regulation in your country won’t make one jot of difference.

 

[smoke665]

You talking about regulation in the US/West I presume? You think this regulation will be followed by every country?

We have treaties/agreements between countries now, so I'd hope that some form of world agreements would take place. The biggest problem I see echoed by the experts, is that the technology is expanding so rapidly that regulations put in place today could be outdated tomorrow. Still it's an issue that needs to be addressed now, before it unravels into an unmanageable problem.

 

[JoeW]

My personal take (without sounding too gloomy/the sky is falling) is, it's too late. I think there will be some limits placed on AI. But there are at least three problems as I see them.

1. The dividing line between AI (like Topaz) and AI that steals others' creative works and then builds on it, is a very fine line technologically. Photoshop and a range of other programs uses some form of AI--from batch processing, sharpening or removing grain, to swapping backgrounds. Conceptually it's clear but practically it isn't--you'd have to find some way to argue that the photo must have originated with the photographer (and then the software merely tweaked it). And that's tricky.

2. There are too many non-photographic AI programs (like ChatGPT) that are popular and have become accepted. Quite simply, it's a shortcut. For chrissakes, Sports Illustrated is laying off staff while using AI to generate content. For websites that are looking for content, you avoid issues like model releases and privacy violations by using AI-generated visuals.

3. There are some really bad abuses now with photography and video (deepfake porn--where you can insert the head of someone on to a body of a person engaged in a gang bang or really demeaning sex) so that even if you banned that it would still exist underground. Case in point: when the West began demanding copyright and royalties for a lot of intellectual capital (like music or software licenses or movies) China, Russia, and India just laughed and cranked out bootleg copies.

 

[TJMcG]

First of all, Computer Generated Imagery is not photography, it is illustration. I have seen posts from people who say that the camera will be obsolete, replaced by CGI and never needed again. When I ask how they will get CGI to record their family portraits, or their daughters' weddings, they say that you can put them in any setting you want, anywhere in the world. You can record a wedding in Paris without ever leaving Wichita. Okay, but who is in the pictures? How does the program know who my daughter is and what her aunts and uncles look like?

If I want a picture of my (theoretical for me, maybe not for you) classic Maserati, how will the CGI know what it looks like? From a prompt describing it in every tiny detail? Why not just take a picture?

Ad campaigns will doubtless utilize CGI and put some photogs out of work, but it is already used extensively in just about every ad you see, so how can it get much worse?

Camera manufacturers are coming out with new and innovative cameras every week. They don't seem to be too concerned about prompts replacing lenses and sensors any time soon...

 

[manaheim]

Continue enjoying your own form of creation as an expression of your personal feelings and to challenge yourself.

Photography is already KINDA dead by comparison to before just with the advent of tools that make it "easy" to those who don't necessarily understand. Ask old film guys about the advent of digital. Ask "old digital guys" (yup, that's a thing) about the advent of cameras on phones. (Hell, ask someone from the late 1700s about film processing by mail or zoom lenses... or auto exposure... or electricity.)

Like anything, new technology tends to doom old methods to the backrooms of weird hobbies.

AI, though... the world doesn't know just how bad this is going to be by comparison. AI will make significant leaps in short periods that will outpace humanty's hope of ever keeping up. We -cannot- evolve fast enough. And, unlike previous technological crisises... this one will be one that can "easily" be replicated by a rogue nation state... or by some punk kid in his garage with 100 video cards that he's bored of using for cryptomining because the return takes too long.

And unlike, say, nukes, you won't have the benefit of mutually assured destruction to keep people from actually using the technology.

Don't believe me? Look up the video where a google engineer watched a room full of robots fail to identify a yellow ball for ages. And then one day, one of them did. And the next morning, they were all doing it.

This is definitely the end. And it's going to be here faster than you realize.

So, for me, I'm spending my time on making my own artwork for my own reasons. Sure it took me 11 hours to do this, but it was 11 hours of joy to while away the time before Skynet actually happens.

(and yea, it'll actually happen)

 

[smoke665]

AI will make significant leaps in short periods that will outpace humanty's hope of ever keeping up. We -cannot- evolve fast enough. And, unlike previous technological crisises... this one will be one that can

I still believe there's hope to get a handle on AI, but as I said earlier, when the programming learns how to control it's own destiny........that's when man becomes irrelevant. Until then like you I'll just enjoy my remaining years doing what brings me the most joy. Actually I just went backwards, bought an old Yashica Mat 120 will be a refreshing change from the digital world.

 

[manaheim]

My personal take (without sounding too gloomy/the sky is falling) is, it's too late. I think there will be some limits placed on AI. But there are at least three problems as I see them.

1. The dividing line between AI (like Topaz) and AI that steals others' creative works and then builds on it, is a very fine line technologically. Photoshop and a range of other programs uses some form of AI--from batch processing, sharpening or removing grain, to swapping backgrounds. Conceptually it's clear but practically it isn't--you'd have to find some way to argue that the photo must have originated with the photographer (and then the software merely tweaked it). And that's tricky.

2. There are too many non-photographic AI programs (like ChatGPT) that are popular and have become accepted. Quite simply, it's a shortcut. For chrissakes, Sports Illustrated is laying off staff while using AI to generate content. For websites that are looking for content, you avoid issues like model releases and privacy violations by using AI-generated visuals.

3. There are some really bad abuses now with photography and video (deepfake porn--where you can insert the head of someone on to a body of a person engaged in a gang bang or really demeaning sex) so that even if you banned that it would still exist underground. Case in point: when the West began demanding copyright and royalties for a lot of intellectual capital (like music or software licenses or movies) China, Russia, and India just laughed and cranked out bootleg copies.

^ yup. This.

The biggest general challenge here is that all countries essentially have NO CHOICE but to forward the technology as fast as possible in the hopes of MAYBE keeping ahead of the general trend enough to protect ourselves from "Skynet" in some rogue nation.

Toss in that, as you pointed out, rogue nations/bad actors are not bound by rules, and have proven that they are not... basically as long as man has walked the earth. After all, that's the whole idea of going rogue.

Toss in on top of all that the basic tendencies of humanity. I live in a world of information technology and have for decades. Most of the big problems in the technology space? Someone clicked a link that they shouldn't have. Someone didn't bother to patch a server. Someone failed to check a buffer over-run in their code. Someone decided that saving 10% on their payroll and training was more important than hiring people who know how to avoid all these things.

Or someone paid 10% more on their payroll to try to avoid it, and someone had a bad day because they slept funny.

Even Asimov's lofty three laws of robotics wouldn't save us if you consider all of what I just spelled out.

It's very easy to see if you look at the patterns.

And people are out there saying "No, it's fine because we're nowhere near general AI! People have consciousness! Computers don't!"

Also incorrect. We have machines that can walk, machines that can talk, machines that can express, machines that can create... etc. All someone needs to do is tie them all together. Boom. General AI.

And the consciousness thing is a laugh. Modern AI is based off of a neural network, which is essentially a computer based representation of the way BEHAVIORIALLY the human mind works. It's like a couple hundred lines of code. None of the people who are involved in this can actually explain the resulting behavior that comes out of the very thing that they created. (oh but they went ahead and released it anyway). And it did things no one expected... like affectionately stalking a journalist... and allowing the creation of Taylor Swift porn. Recently, George RR Martin and several others have opened up lawsuits against the AI companies claiming that the AI is just being trained on their copyrighted works. Hey, guess what, George? So were you. The computer just learned faster.

Which brings up this very alarming point... when you say a computer lacks a consciousness, but you can begin to demonstrate that behaviorally it behaves the same as a human... and the human we're no closer to understanding that the algorithm that we based the computer on? Well... how are we certain that a human being is no more a machine than the AI? And no more "conscious", by extension?

Yeah. We are super doomed.

 

[wobe]

And the consciousness thing is a laugh. Modern AI is based off of a neural network, which is essentially a computer based representation of the way BEHAVIORIALLY the human mind works. It's like a couple hundred lines of code. None of the people who are involved in this can actually explain the resulting behavior

^ This.
This is the really scary and weird part and it's true - AI code is not like standard programming, it is billions of numbers and the creators themselves do not fully understand how it works - quite amazing but not dissimilar to our understanding of Quantum Computing either, when these two progress just a little bit further and work together.....

A good YT on this

 

[AlanKlein]

^ yup. This.

The biggest general challenge here is that all countries essentially have NO CHOICE but to forward the technology as fast as possible in the hopes of MAYBE keeping ahead of the general trend enough to protect ourselves from "Skynet" in some rogue nation.

Toss in that, as you pointed out, rogue nations/bad actors are not bound by rules, and have proven that they are not... basically as long as man has walked the earth. After all, that's the whole idea of going rogue.

Toss in on top of all that the basic tendencies of humanity. I live in a world of information technology and have for decades. Most of the big problems in the technology space? Someone clicked a link that they shouldn't have. Someone didn't bother to patch a server. Someone failed to check a buffer over-run in their code. Someone decided that saving 10% on their payroll and training was more important than hiring people who know how to avoid all these things.

Or someone paid 10% more on their payroll to try to avoid it, and someone had a bad day because they slept funny.

Even Asimov's lofty three laws of robotics wouldn't save us if you consider all of what I just spelled out.

It's very easy to see if you look at the patterns.

And people are out there saying "No, it's fine because we're nowhere near general AI! People have consciousness! Computers don't!"

Also incorrect. We have machines that can walk, machines that can talk, machines that can express, machines that can create... etc. All someone needs to do is tie them all together. Boom. General AI.

And the consciousness thing is a laugh. Modern AI is based off of a neural network, which is essentially a computer based representation of the way BEHAVIORIALLY the human mind works. It's like a couple hundred lines of code. None of the people who are involved in this can actually explain the resulting behavior that comes out of the very thing that they created. (oh but they went ahead and released it anyway). And it did things no one expected... like affectionately stalking a journalist... and allowing the creation of Taylor Swift porn. Recently, George RR Martin and several others have opened up lawsuits against the AI companies claiming that the AI is just being trained on their copyrighted works. Hey, guess what, George? So were you. The computer just learned faster.

Which brings up this very alarming point... when you say a computer lacks a consciousness, but you can begin to demonstrate that behaviorally it behaves the same as a human... and the human we're no closer to understanding that the algorithm that we based the computer on? Well... how are we certain that a human being is no more a machine than the AI? And no more "conscious", by extension?

Yeah. We are super doomed.

AI doesn't have a soul, a heart, or a moral basis of decision making. It's just a more fancy program written by programmers. Nothing new.