IMO not really. We’re making the decisions, not pushing a “meta-button” and letting machine learning take it from there to completion. Darktable has no AI that I’m aware of. It’s just a tool to do what we tell it.
Also it’s not the AI technology, per se, but rather the use and (potential for) massive misuse of it over the long run, as well as the dishonesty that will inevitably be involved. We can be dishonest now but AI multiples that possibility by orders of magnitude.
100% agreed. I believe people will really “wake up” to its dangers in the next few years. Voice synthesis is almost indistinguishable, video deep fakes are also getting there. Image is still behind in the realism front but otherwise it’s evolving at a scary rate. And this is only what’s public, imagine what governments like the US’s already have behind closed doors.
Not to be a doomer but it’s hard to look at it in a good light given how it’s gonna be misused.
This is a decent article based on Nick Bostrom’s book.
I wonder how many of those websites one finds that appear to be nothing more than a container for ads with semi-plausible waffle filling out the spaces are actually done that way.
Beyond the (non-)AI illuminant selection, there are two tools integrated via Lua scripts that I’m aware of, both related to tagging, not image processing.
But anyway, I think we need to differentiate between AI used to generate content that’s not art (replacing stock photos, creating wedding photos, rendering portraits), and AI as an artist’s tool. Whether it’s a picture or text, as long as the intention is that of self-expression, evoking thoughts and emotions, a human directing an AI is still an artistic process, at least in my eyes.
An example: AI sky replacement may be simply faking something to make it ‘prettier’, or it could be used to create an image that’s full of tension because the sky is so out of place, contradicting the rest of the scene.
I enjoy creating images with Stable Diffusion and Midjourney, but to me, that’s a very much different thing to taking photos, especially if the photos are documentary.
I’m currently participating in a “100 bird species in one year challenge”, and creating a rendered image of a bird would make absolutely no sense in this context. Likewise, I was asked to take photos in my brother-in-law’s mother’s funeral at the end of the last year. Replacing or adding/removing arbitrary details to them surely isn’t something I’d want to do.
Holy words. I imagined that my thought would inflame spirits, but rightly so. We are facing a new and disruptive technology which, like all new technologies, will bring benefits and misfortunes.
I only bought my first smartphone 3 years ago because I didn’t want to become a slave (and I only bought it because there was Covid and I didn’t want to end up in intensive care without a minimum of contact with my family).
I waged war on PCs as long as I could, then realized it was a losing battle, and made PCs my job.
I didn’t want the microwave oven, but I gave in there too.
I’ve had battles against street webcams, but I’ve lost.
I didn’t give in to Microsoft or Apple, but I finally gave in to Google.
I didn’t want the AC, but I had to succumb.
I still don’t have a dishwasher and I sold the car.
AI is here to stay. Is it good? No. Is it bad? I don’t know. Certainly it can be put to good use.
Ironically, I have been toying with AI models lately out of curiosity, just to see what’s currently possible. I can run the audio through the Whisper speech transcription engine and get you a timestamped transcript pretty easily.
Digikam uses facial recognition “learn” faces from the people that you have tagged and tries to find more photos of them so you can tag the correct name on a photo. It sucks. I have often wondered if there’s a way to make a Digikam plugin and use one of the up-and-coming AI facial recognition models to train the model on face tags already in the Digikam database and do a better job at making face tag suggestions.
Finally, if you want a laugh, download the tiny little program photils-cli and feed some of your pictures to it to see how it classifies them. It uses a trained neural network for image recognition.
Is AI going to demotivate or even dilettante photographers? No chance. Has it done so for chess players, mathmeticians, model makers?
Besides, until it grows legs and develops genuine emotions, it will never be able to do the most important thing of all — the one thing that makes great photographs (and great photographers) great: be there, both physically and emotionally, to capture that powerful or beutiful or humerous or desisive moment.
And if it ever DOES manage both, then I guess we could drop the ‘artificial’ bit and just call it intelligence — and if THAT happens, Robby the Robot stealing my camera will be the very least of my concerns!
Sorry, @snibgo; I meant to post a general reply rather than a reply to you directly.
It’s either going to turn everything on its head or be a weird fad everyone forgets about and we’re all sitting around in 20 years going “remember those weird chat bot and image generator things?”
It’s either the smartphone or Pogs, nothing in between.
I think, at least as photography goes, this is going to be something akin to the HDR clown fart image craze of the mid-2000s. Everyone did it (don’t lie to me) and it looked new an interesting but it eventually evolved into something more useful and less cartoonish. So far these generative ML made images have a specific look to them like that and it may be harder to shake that than we think. However I can see them growing in usefulness for very specific tasks. Sort of like what that Luminar thing does with the sky replacement already.
A lot of product, food and stock photography is on the way out regardless anyway. 3D rendering has become so good a lot of places just pay someone to render a model of the thing they’re selling instead of using a photographer.