Your Answers to RPS AI Questions

Which of the following best describes your view of AI for photography? (max 2 options)

  • An opportunity

  • Cautious

  • Dangerous

  • Excited

  • Fearful

  • Couldn't give a monkeys (added for Dave)


Results are only viewable after voting.
Messages
8,318
Name
Ian
Edit My Images
No
I just received a survey from the RPS as they're going to be having an AI conference on the 9th October. Sadly the survey only allows poll type answers, so here are the questions in case you have an opinion. I'm interested to hear people's thoughts, both from the point of view of an amateur through to professionals. I was a bit disappointed at not allowing for any explanation, and the questions felt a bit too generic to get useful answers, but I was interested in what people might think. I can only put one poll in a post so I picked the last one.

1. Do we need traditional photography if images created by AI are better and quicker to create than photographs taken by people?
This had a yes/no/maybe. My answer was yes. As far as I'm aware, AI has no concept of what creative vision is. Let's face it, most humans don't know what it is. And how can anyone assume AI photographs will be "better" than human created ones.

2. Do you consider AI created images to be real photography?
Another yes/no/maybe. I said no. Photography for me is about using a camera. Regardless of the definition of the word, that's what it means to me. It's not a photograph, it's an illustration. I took this to mean full images created by AI, rather than someone doing a sky replace on their own photo. To those, I'd answer "maybe" only because "sort of" wasn't an option.

3. Is it fair for AI algorithms to be trained using images without permission or payment for the original creator?
No. But if people put things online, there is always the possibility other people will nick it. Doesn't make it fair, but it is what it is.

4. Could AI imagery lead to a rise in fake news and lying?
Yes. Of course it could. Will it? Who knows. But I do think most people are lazy, and if they can get traction on social media, they'll take the easiest route to get there. "Look at this mouse on a railway line." or "The aliens are coming!" The Sunday Sport was doing this in the 80s!

5. Which of the following best describes your view of AI for photography?
I put this one up as a poll because I thought it might generate the most interesting results.
I went for cautious and opportunity. For me, AI photography is really useful for game modding and illustration. I'd personally never use it for actual photography, simply because it doesn't interest me. I think that some careers will take a big hit (illustration & advertising) but others will still be needed and AI will make jobs easier & quicker at the editing stage (wedding, sport). Some will make people suspicious (nature, reportage) and things like fine art landscape will probably take a big hit. Architecture & Real estate will probably heavily use it to modify real world original images by the photographer. I also think that using AI to make illustrations using real photographs will be another skill/career path that will evolve.

All in all, it felt like a weird bunch of questions to get a whole conference started. Q1 felt a bit daft coming from the RPS. I think Q2 is going to be discussed to try and get a handle on what the RPS will/won't allow for submissions in future, and I could see any photo competitions having the same problem. It did make me smirk to think the only way they could prove a photo was unedited would be to see a physical negative. As far as Q3 goes, the issue of pinching images has been around since the Internet so I don't think it is something specific to AI per se. And Q4 felt like a bit of a dumb one unless I'm just overly cynical.

Anyway, poll's up. Questions are above if you want to answer any of them. Like I said, I'd be interested to hear the results. All the discussions around Photography & AI that I've seen have been people full of fear. It would be good to have a reasoned discussion. Or not :)
 
There is one option that I immediately thought I would tick but it wasn't there and that is -- really not bothered. Let me clarify, as that may, at first, seem rude: I can see it may well be very important for members of the Royal Photographic Society with professionals whose livelihood depends on their photography but here on this forum, where most photography is done for personal pleasure, AI is probably not going to have much of an impact. I mean, we've moved from B&W to colour, film to digital, SLRs to mirrorless and technology has marched on with ever greater possibilities, but there are people still using B&W film, still buying SLRs and still switching their cameras to fully manual. From the amateur photographers point of view, I can't see anyone is going to be that bothered about AI. It's going to come, it's just something that can't be stopped and, let's face it, it has been getting harder to be sure what is real and what is artificial for some considerable time, so I suppose if we could be sure that something was generated in a computer it might be rather reassuring as we can now assume everything is fake and work around it.
 
Interesting poll.

1. We need traditional photography to feed the AI monster. I believe it works by searching around for pictures to use as fodder. Cars and peoples costume change, so I guess it needs new photography.

We need traditionnel photography to show actual real things. Reportage for example needs actual pictures.


2. Easy and not so easy to answer. No, it is not photography, it is illustration. I know our photography strays into the illustration category, when we use the clone tool or exaggerated HDR.

3. No, but it is just part of the "sharing" or rather "stealing" economy. I guess at some point there will a copyright infringement case, to spoil the party. I have given up worrying about people stealing my pictures.

4. Yes, it will, just like the Soviets "disappeared" people from film based pictures. Fake news just got easier.

5. It will have a big impact on professional photography/illustrazioni. A lot of product "photographs", that we see are already computer simulations. Cleaning up architectural and fashion shots will get a whole lot easier, and I think creating effects and improvements in wedding photographs will enter a new era.

I am already seeing some amateur photographers using it to create effects, grass lawns, paving Oxford Street or a lake in front of St Pauls , gives the idea.

It could be a useful tool for cleaning up pictures. Interior shots where a large window in the scene causes contrast killing veiling flare comes to mind. It would be really useful in this case.

To conclude, it does not bother me too much as I like the challenge of making the best photograph I am capable of making of a given scene. Making fake photographs does not interest me, as I know it is fake and pointless. It could be useful to correct lens errors and for minor retouching.
 
I think @NCV Nigel made the answers I would, but in summary: as above, it provides opportunities for creativity and entertainment for enthusiasts, but threatens the livelihood of some professionals.
As to AI needing traditionally-created images to feed the algorithms, no; the bigger the data on which the algorithms can work, the better, however in the absence of traditional images, AI will feed upon it's own, to create increasingly mutated versions of its own reality. It's the nature of the beast. So if we all stopped taking images as we do, and posting them online, the AI bots would in due course start producing some very unreal views of people and scenes, which would be mainly amusing but somewhat frightening too if misused.
 
What's the difference to most people between what AI is today, and sophisticated editing software when it was a new concept?

I remember similar things being said then, but now how many digital photograpehers do not do some kind of editing/adjustment?
 
What's the difference to most people between what AI is today, and sophisticated editing software when it was a new concept?

I remember similar things being said then, but now how many digital photograpehers do not do some kind of editing/adjustment?
Not just digital photographers!
Dodge/ burn / curves spotting is easy on a film scan have been easier than it was in a full wet process print for a long time. Now Ai makes a great job with content changes / object deletes on a scan, matching tones and grain structure very well.
Just another tool in the box.
 
really not bothered. Let me clarify, as that may, at first, seem rude:
Not rude at all. It's your honest opinion.
Where's the 'Couldn't give a toss' option? :)
I fixed the poll. Just for you :)

I guess my only worry was that the perceived notion of what is a good photograph today will change. But as Martin says, those who are doing photography (without AI) for pleasure will continue to do so. It'll just be another thing.
 
Last edited:
Not just digital photographers!
Dodge/ burn / curves spotting is easy on a film scan have been easier than it was in a full wet process print for a long time. Now Ai makes a great job with content changes / object deletes on a scan, matching tones and grain structure very well.
Just another tool in the box.
I count that as digital, I also scan negatives and process them digitally :)
 
  • Like
Reactions: zx9
One very important aspect of AI is: just how many lies will it be used to tell?

In 1966, the English science fiction writer Phillip E High postulated a world dominated by what we would now describe as AI generated illusions. ( https://mporcius.blogspot.com/2016/03/the-mad-metropolis-by-philip-e-high.html )

In this story, the inability to tell truth from lie has created a world in which the biggest liar wins and the person who sees through the lies is public enemy number one. While (inevitably) naive in some ways, this book seems to me to be now more relevant than when it was published.
 
One very important aspect of AI is: just how many lies will it be used to tell?

In 1966, the English science fiction writer Phillip E High postulated a world dominated by what we would now describe as AI generated illusions. ( https://mporcius.blogspot.com/2016/03/the-mad-metropolis-by-philip-e-high.html )

In this story, the inability to tell truth from lie has created a world in which the biggest liar wins and the person who sees through the lies is public enemy number one. While (inevitably) naive in some ways, this book seems to me to be now more relevant than when it was published.
Rather like Animal Farm :)
 
One very important aspect of AI is: just how many lies will it be used to tell?

In 1966, the English science fiction writer Phillip E High postulated a world dominated by what we would now describe as AI generated illusions. ( https://mporcius.blogspot.com/2016/03/the-mad-metropolis-by-philip-e-high.html )

In this story, the inability to tell truth from lie has created a world in which the biggest liar wins and the person who sees through the lies is public enemy number one. While (inevitably) naive in some ways, this book seems to me to be now more relevant than when it was published.
Rather like Animal Farm :)
Or Donald Trump or Boris Johnson?
Back on topic, I agree that the questions are vague and often difficult to answer with a yes/no. It seems to be very common practice with all sorts of surveys and public consultations and I don't know whether it's due to incompetence or something a bit more left-handed, possibly guiding people towards what they want to hear and/or stopping them from providing their own views.

As for me, AI is what it is, it's here to stay and although it has clear benefits it has clear dangers too, not least of which is due to the fact that it will be impossible to control or police.
 
While (inevitably) naive in some ways, this book seems to me to be now more relevant than when it was published.
Like much science fiction with its dystopian predictions?
AI is what it is, it's here to stay and although it has clear benefits it has clear dangers too, not least of which is due to the fact that it will be impossible to control or police.
Much like the internet, then ...
 
Much like the internet, then ...
Yes and no.
Yes, because fake info is all over the internet, and its success relies on the stupidity of the people who accept postings at face value.

And no, because fact-checking is usually easy for those who bother to do it, just as fact-checking can easily be applied to stories in newspapers.

There has always been manipulation, I remember an (unimportant) one where a government minister was supposed to turn up at a school but didn't, so the photographer took the usual smiley group photo without him and added him in later.
And there was the North Korean rocket launch too, faked in Photoshop so badly that my cat could have done a better job if I had a cat . . .
And, pre-digital, there were the famous Russian political photos where someone who had fallen out of favour had been expertly and literally airbrushed out.

AI is very different, because it's already so good that it's extremely hard to spot. I read, a couple of years ago now, about AI that can effectively turn clothed people into nudes. To try it out, I used a photo of someone wearing normal clothes, and in just a few seconds it came back with a photo of her nude, and the worrying thing about that was its accuracy, which I knew to be incredibly good.
 
Fakery has been around almost as long as photography/image creation.

For example, one of the most well-known images of Lincoln isn't his body (although it's not a photo either).


But at least that had some sort of human input. One of the biggest issues I've experienced with using AI is there's no real control. If you want to put the same person in 5 different scenarios and use the same key words for the person, but a different set for the background it will generate 5 different people. This will change. But as a shortcut to making a storyboard for example, it doesn't really work.

I know Getty and/or other libraries will no longer accept any AI generated imagery due to the issues of copyright. But we've had a Firefly presentation at work from one of the developers at Adobe. They way their generative AI works is by ONLY using Adobe Stock as its source material, so neatly sidesteps the copyright issue there. Any images you create using Firefly will have all the original credits for the source imagery in the metadata, so if they're used in advertising work (for example) the original creator will still be compensated.

As for whether or not it's photography, for me the answer is fairly simple. No.
 
Last edited:
Some artists are using AI to create interesting images. Someone I know vaguely online has been using MidJourney to create some images and is very excited about the possibilities to create a photo realistic images. See https://www.lensculture.com/article...ca-ai-generated-photos-from-the-1940s-and-50s

However, I can also see that for some sectors it could almost decimate their work, in the same way that Copywriters are also worried about the text generative capabilities.
 
For those of us who do photography for pleasure and compete in club, national and or international competitions, AI generated images will not be allowed. A significant problem is that there are many photographers who produce excellent composites which will come under suspicion unfairly in future.

Dave
 
I fixed the poll. Just for you :)
Thank you, and voted. (y)

AI is here to stay. Pointless moaning about it. Photographers of all ilks will just have to get used to it. Some will use it, some will shout abuse at it, others will ignore it. The world will keep on turning and photography will carry on evolving. FFS some people still make tintypes, never mind shoot film! :D
 
Back
Top