« Camera Chooser | Main | What Would You Do If You Had Money? »

Wednesday, 19 April 2023


Feed You can follow this conversation by subscribing to the comment feed for this post.

The modern version (GPS tracked in this case) of your running and cheating story is in the Guardian today. https://www.theguardian.com/sport/2023/apr/19/scottish-ultra-marathon-runner-joasia-zakrzewski-disqualified-after-using-car-in-race

A good one. No matter if it's true or false.
BR, Yoram

You cheeky monkey you!

I’m glad he submitted the AI- it helped further a much needed conversation into all things fake these days.

Good to know all three of you were considerably more honest than Rosie Ruiz…

From Guardian article: 'A spokesperson for the World Photography Organisation said Eldagsen had confirmed the “co-creation” of the image using AI to them before he was announced as the winner'.

I read that to mean they knew, before awarding him the prize. So is more to this than is visible at first.

Especially think this is true as very slight inspection of the image (version on his site is better) shows it to be not real: person's right hand on shoulder is very strange, especially tip of index finger, wrist of that hand is also odd; same person's partly-hidden left eye and cheek are weird, their ear is ... what I do not know.

Think is clear judges were in on this at least in part.

You may have read in The Guardian that it has found there are fake AI created “Guardian articles” being posted that they have had great difficulty establishing that that are indeed fake and that even the (genuine) writers could not be sure they hadn’t written them. This has the potential to be a more serious problem than fake photos in competitions.

When will AI be used as a judge?

Here's a faker who DID finish and was surprised to find herself the winner. She which would have won (in 1980) with "the fastest female time in Boston Marathon history as well as the third-fastest female time ever recorded in any marathon":
The article ends with ""she jumped out of the crowd [half a mile from the finish], not knowing that the first woman hadn't gone by yet. Believe me, she was as shocked as anyone when she came in first."
Note: This comment did not employ any AI in its writing - just HI.

I think there was a statement from the contest to the effect that they knew the entry was AI-assisted but accepted it because the category allowed it. That too might have been an attempt at face saving but I’ll see if i can find the quote. I don’t know why you won’t accept the idea that it was both a trick and a high minded prank. Do you think no one learned anything from this or became aware of AI fakes because of the incident?

I think the fingers are a dead giveaway. AI doesn't seem to be able to do them right yet. Shame on the judges for missing that.

I think everybody in the AI business would have been able to tell at a glance. The woman in the back and her hands look unreal. The hands always give it away ---so far. Who knows what will happen in a year.

Another interesting thing about the generated images is that they all look hyperreal and photoshopped ---kind of like 99% of images on the internet, which were presumably used to train the models.

The "cheeky prank" can usefully function as a wake-up call about the overwhelmingly repetitive nature of so many "award-winning" images in the digital age.

Large language AI models basically converge upon, and reiterate, what's already the most popular on the Internet to the point of extreme cliche'.

That may be the most embarrassing aspect for the Sony judges - they chose of a reiteration of existing cliches. Serious art photography ostensibly is about developing something new, different, and hopefully better, not repeating cliches.

To be sure, if the entrant only confessed to the falsification after being given the award or only after falling under suspicion, then that of course would have been evidence of a dishonest intent.

One could argue that an image created with a mirrorless camera that uses AI a driven scene mode, auto focus or esposure algorithm that has software defined lens corrections is fundamentally an AI created image :)
But, but, but... No buts
It's actually one of the reasons I have stayed with Pentax even though I work with technology as my day job - there is still control available to you the user, photographer, artist. Wysiwyg is still in play for the most part, the lens is optically corrected and doesn't need firmware to work, you have to specify where to try to focus ;), there is a little indicator to tell you how your exposure is looking.

After prep school you have divulged that you went to Dartmouth. What courses there did you particularly like? English, we might assume, but what else?

[Art, art, art! That's why I got in to Dartmouth--for the class of 1979 the Admissions Department was being friendly to students with artistic promise. --Mike]

I wonder what the instructions to the AI were? "Create a photograph that will win the Sony World Photography Awards".

I think we are reaching a moment similar to the following.
It is 18800 in a Paris cafe and a group of artists are sitting around brooding over their absinthes. Renoir says "Have you seen photography? What are we going to do about it"? Monet answers, "We need to do something that cameras can't do. What about this for an idea? Let's just do impressions."
Conclusion. A! will totally take over digital image making. It is already doing it in the processing stage. This trend wont be reversed. Perhaps "photography? will come to mean something where provenance can be establishe because the image is made with negative or positive film and images are prints are made from the film via an enlarger or contact process.

One thing that struck me about the Guardian article by Zoe Williams was the description of the image. "the older one behind with her hands, weathered to the point of being misshapen, on the younger woman’s shoulders." Weird-looking hands is one of the dead giveaways for an AI image. I suspect the author has not seen too many AI images. The weird hands phenomena will probably disappear in a few weeks given the speed that AI learns.

"Personally, of course, I would rather hear what the judges of the contest have to say for themselves", maybe they were AI too!

"The Guardian (whose journalism I support, by the way)" -- me too! I greatly appreciate reading a "newspaper" that I know I can trust. Sure, it leans left, but so do I.

Re the preferred camera in the next post, although I'm a Pentaxian (K-5), the monochrome K-3 III is of no interest to me. I was out driving yesterday, late afternoon in a burning-off area and the sky was hazy and the shadows blue-ish. Suddenly I saw a white bulk oil tank with a red logo. The combination of soft hazy white, blue-ish shadows and red logo would have made a marvellous picture. But only in colour. In greyscale it might have been nice, but nowhere near as nice as in colour.

My point is, I see colour and if I had a monochrome camera, I would *have* to carry and use a full colour camera first. That would mean always carrying two cameras.

So I missed the shot yesterday. I'm out of the habit of always taking my cameras these days and anyway, the traffic was too difficult. The one that got away.

PS: it was the eclipse this morning, totality at about 11.30am here. My lawnmower guy asked me if I took photos. Nah, what of? It got very dim and quiet here in Perth, but there wasn't much else to see. Seen one eclipse, (in my 76 years) seen 'em all. Next time.

I recently had a guy commenting under one of my portraits in a huge portrait group over on Facebook, accusing me of using AI to create it. Thing is, that portrait was taken back in 2013.
There's a huge disruptive power in AI, and it's only the beginning, too. We're all funked, and no mistake.


One of the great injustices in life is that we didn’t all have a houndstooth-wearing, Mercedes-driving friend who smoked pot from a briar pipe.

I think there is going to be a lot of this. For myself, I am more interested in the delight of seeing new things than worrying too much about what kinds of intelligence produced them.

But the implications of this are going to be difficult to grapple with on the monetary side of things. Who's going to get the money from selling a song, say, in the "style" of James Taylor, or JayZ? Can an AI-generated image be copied and pasted into an ad campaign like something from a "creative commons" license? What will this do to stock photography as a business? Or the concept of "Fine Art"?

And there is also the issue of our finite attention. If AI is making the best pop music, photography, or writing the best novels, will human artists be able to make a living doing those things? It is pretty hard already for those folks.

Whoa, Nelly. Here we go.

FOX must be salivating.

I work with a company in Scotland that produces fine art prints and merchandise based on original artwork

We are looking at assuring buyers when products as based on work "produced by the human hand"

A bit like the "Appellation d'origine contrôlée
" or in the UK the old Fine Art Guild stamp (https://www.fineart.co.uk/buying/stamp_of_authenticity.aspx)

On review, I not only eat my words, I see a far worse tragicomedy than even Mike's characterization.

What I understand from the Guardian article is that according to the contest organizers, the category was open to involvement of AI (along with just about any kind of technique and pre- and post-manipulation) and that they'd been eager to work with the winner to use the award as a platform for discussion about AI issues in photography.

That much I recalled. And, by the by, it makes the photographer's claimed purpose of showing that contests aren't ready to deal with AI entries a bit silly.

But the worst part is that had the photographer, say, fixed the hands or changed the tint, based on his "wealth of photographic knowledge" as he'd represented to the committee--in other words, if he'd gone all in on the "cheat"--he may not have been a cheat at all per the category rules. He might have been able to legitimately accept the prize AND had an international platform to launch and steer the discussion he claims to want for the community. At worst, it would have been a legitimate chicken-egg type controversy.

Instead, he basically yelled "gotcha" after a lame prank and ran away. It's as if he'd impersonated a clown at a clown convention and won "Best Makeup".

So problem for me is not that he "cheated", but that the "cheat" resulted in a precisely pointless and counterproductive self-own (if you take each party's statement at face value). In other words, this contest winner could not be more of a loser than he's made of himself. And having lost his nerve he's robbed us of a possibly worthwhile discussion.

Lost in all this is the question of what we want from a moody photograph like this and why we care how it's achieved. Are we after some kind of emotional "truth"? Is it somehow better if the people in that picture are models lit by a pro and the shot was the best take out of twenty? What if they are ordinary people with a relationship but it took thirty takes and the photo was significantly enhanced in Photoshop? By an AI?

I agree with Dave Millier's response. This article is a work of art. This is what makes TOP THE Online Photographer.

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007