This from my friend Joe Kashi, who was active with the ABA Law Practice Management Section (now renamed), which was then the primary locus of law office technology in the ABA. He is also a fine arts photographer.

=========

I am sending along some thoughts that I wrote about the intersection of traditional photography and the many AI tools, both as secondary adjunct tools like better noise control, and primary creating of images by generative AI.

We all use a variety of tools on that slippery slope. Where to draw the line on AI ? “It depends.”

It’s easy for evidence and photojournalism – no changes beyond what the lens sees, maybe with tracked, reversible corrections to overly dark or light areas to bring out data, but no change to fundamental data recordation. And, corrections easy to track in standardized formats. That one is pretty easy.

For evidence and journalism uses, there are now a variety of software tools meant to authenticate images as physically generated recordations of fact and not AI-manipulated . These can enhance reliability in this context. Nothing’s truly exact. Everything in experience, including physical science, is something of an abstract interpretation. JPEG itself is a generally agreed-upon interpretation that we accept as generally describing a bigger “reality”.

Fine art photography is definitely further down the slippery slope where pretty much anything is acceptable so long as a worthy image results but, to my thinking, with two caveats.

For many photographs, especially those that are landscape and street photos, I suspect that much of the audience reaction to a photograph, especially those of people and landscapes, may often depend upon whether the photo seems to comport with the viewer’s own framework and memory of visual experience and the experience of a place. The optimum point and validity for such images is obviously pretty personal and subjective for each viewer.

The second caveat might be that audiences may still prefer a photograph, especially a landscape or people or street photograph that gives a strong sense that the photographer was actually physically present and experiencing that which is depicted.

Tastes probably change and people like me will likely be obsolete for a while as society becomes “trained”, just like AI LLMs, to prefer glossy AI-generated images, just as we have seen with a lot of people preferring photos on aluminum, etc.

In the long run, though, those pendulums of taste keep swinging back and forth. Years ago, there was so much abstract photography that I recall urging more focus upon realism. Now, it’s swung quite far in the other direction, at least if we look at “high-faluting” taste-setters like Aperture.

I recall being in a commercial photo gallery in Seattle about 2019 where the owner told me that she would only carry traditional process photos, i.e., silver-gelatin, Platinum process, etc. That struck me as being a bit extreme by then, but she was representing people like Michael Kenna, so it then appeared to be a workable niche for her. Now, though, it’s all the avant garde rage again.

So, for fine art photography, I think that you can make a strong case for using whatever AI tools you might like, but as adjunct tools rather than as a mechanism for primary creation, just so long as a sense of place, experience, and the photographer’s experience remains and shines through.

[Ed. Note: there are technical solutions to ensuring that video has not been tampered with; more of that anon. It would be a shame if we were left with only eyewitness testimony, since eyewitness testimony is the word testimony there is.]

Faking A Shot For News Or The Law

There are a lot of ways in which faking can be suggested but efficiency does still come down to trusting the source.  It’s too slow and resource-consuming to have to do otherwise in a modern high-speed news cycle.  

Some camera makers, such as Sony and Nikon, are exploring how to put an invisible and unalterable watermark into digital files taken by their higher-end cameras used by professional.  This is probably the cleanest approach for authenticating evidence and journalism but not likely to be widespread in the immediate future.  Realistically, this should be doable by all camera and cell phone makers.