A lot has been already mentioned in this thread. The replies also show that AI is a wide variety of things.
Where it comes to using other peoples creations/ photos without proper credits, I considered it 'wrong'. But that is largely the case for situations where things get added to a scene, the so-called generative AI. That said, the AI is trained on billions of images and generating things is not a mere copy-paste from another persons photo, but it is the result of analysing many photos to see a pattern on pixel levels and then using the knowlegde of those pattern to create something. So crediting original creators is merely impossible.
There are several things that are now labeled as "AI" that were around before AI became a thing.
Forms of automation that exists for longer time, were in fact already 'artificial intelligence' . Making selections based on luminosity levels or color ranges and changing brightness, hue etcetera, already was a form of AI. i.e. The computer analyzing the photo, determining a value and determining to include a pixel or not. Then when we as user move a slider like a hue slider, the computer determines how to alter the included pixels, but not necessarily alters each in the exact same way.
The blend modes in programs like photoshop have intelligence driving them. With a 'color dodge' blend mode for example, the software has the intelligence to know how to change the appearance on a pixel level based on how things overlay. The analogy of us as people, would be tracing paper with different levels of transparency with different colors in different areas that in the end combine to a desired end result.
Automated focus stacking is based on the computer determining what is sharp in a photo, in which photo parts are the sharpest and then masking out areas on photos from the stack that should not be used. We as people have the 'intelligence' to see what is sharp, and determine what to mask, but the computer just is faster in doing so.
Subject detection masks are an addition that take over the manual labor, but it is not (strictly) based on pixel properties like a hue and brightness and they are trained on many photos (of others). They do however take away manual labor and do not add anything based on the work of others. So would credits to the creators of the photos the AI was trained on be 'required'? (Apart from the practicality that crediting is shear impossible).
When using automatic masking ones life gets easier as you do not have to do it by hand, but the end result of the post-processed photo is the same, but just accomplished in a shorter amount of time. So from a 'photography' point of view it should be allowable for sure.
I have nothing against compositing, but the 'art' is then 'compositing' and not 'photography'. Photography just becomes a mean to achieve the goal, the composite. And although compositing relied heavily on photography and the art of shooting for composite purposes, the advancements in AI do change compositing a bit more. Not only easing task that used to be manual/ labor intensive, but also the option to generate more rather than needing proper base photos.
For small removal tasks I think AI is fine, like it was mentioned earlier in this thread it used to be done by hand through cloning and stamping and the like. Also noise reduction through AI is fine in my opinion, since it removes something that wasn't here in the first place but was induced by the sensor. In both cases the essence of the photo remains and I still consider it 'photography'. When removing things that were in the photo, but did not necessarily be in the photo, that is fine with me. For example removing people from a scene, that in theory could have been photographed without the people is fine with me (it is not practicaly possible to be at a Venice square alone without other tourists, but in theory it is achievable).
So for me use of AI is allowed for 'photography' and the 'photos' on this forum. In some cases the use of more impactful AI could be fine, but then it should be mentioned and preferably should have resulted in a photo that in theory could have been taken in camera if one would have had control over the circumstances. When it goes towards another art form that 'photography' it should not be on this forum, since the forum is about photography and not on other forms of art.
To conclude there are other forms of AI that I think were not mentioned in this thread and that has less to do with adding stuff to a photo or removing parts. The ones I can think of are:
- AI (assisted) culling . For some genres this AI can be really useful and although trained on many images, it does not directly makes use of other peoples content. (It is shear impossible to credit all the creators who's content was used to train the culling AI). I have looked into automated culling, but for my purposes and style of culling, it is just not there yet (my shooting being sports mostly).
- AI editing (basic). A program like Aftershoot can be used to do basic edits to your photos. Either based on your own edited images or a style that was learned from other peoples photos. I fed it with a series of my own photos and use the style to a basic edit of my RAW photos. I feed new RAWs to the program and it analyses them and determined the adjustments I likely would made and then I import the edit settings into Lightroom. It is like a preset in Lightroom, but rather than setting each value the same for each slider on each photo, the slider values are tweaked per photo. Also some straightening is done. Where culling does not fit my needs, the edits in >95% of the cases are pretty much like I would have set it, with some photos needing some minor tweaks on some of the sliders.
- AI retouching. I have not used this, as I hardly retouch photos. But for wedding, event, portrait photography I can see the benefits of it. As far as I know it can do retouching based on you style or if you wish so the style of other photographers who made 'their style' available. It merely are retouching edits that you would other wise have to do manually in Photoshop (or similar programs). Things like skin softening.
These three applications of AI can be used to automate things you would otherwise do manually and will not result in a different outcome than when you would have done things manually. This form of AI can be pretty useful (i.e. time saving) and still rely on the skills of the photographer.
) ive been on Linkdin for about a year or more. looking for commercial photo work. i get INUNDATED with 'job openings' (not necessarily 'offers') for people to join some team to use AI to help something. i dont even bother reading past the AI part of the job descriptions.
