I learned both a word & a concept today --
indexical.
Oh, there's no bottom to that pit.
I could have said, ala Barthes, that one cannot predict which detail in the photograph will become a "punctum" for the viewer, carrying some importance of meaning that subjectively extends beyond the objective appearance. "I remember the scar on the chin of my great-great-grandfather--everyone in the family talked about it and blah, blah, blah". But here's this photo that doesn't show it, because the AI engine though it was a scratch on the tin-type or something, or it didn't feature on the face that the AI engine was morphing to have the same shape as in the damaged photo. Maybe the AI engine wasn't a party to those bits of family lore that brought it extra meaning. Thus, the indexicality of the photo would be needed for the viewer to say, "There's the famous scar! The altercation that caused must have been before ____!"
Yes, that's a silly example, but it illustrates the point.
Barthes also talked about the "studium"--the objective facts presented in a photograph that describe its social, cultural, and historical context. AI that uses replacement strategies can mess that up, too, but it's less likely.
In semiotics, the relationship between a sign and the thing to which the sign refers is the indexical relationship, and photography enforces that relationship by the reality that the only marks on the photograph (the sign) were made by light emanating from the subject (the thing to which the sign refers). (Digital photography processes didn't change this, but it sure made it easier to mess it up.) That's what makes photography unique as an art form.
This is all important to me both as a photographer and as one who has carefully studied the relatively few photos that exist from my own family's past. I've restored a lot of old photos in my day, having established my digital darkroom all the way back to 1999. Keeping it real requires expertise to know what's photo and what isn't, and AI might be quite good at that. Or not. But as soon as it tries to fill in the blanks, the indexical relationship is lost.
Of course, AI isn't the only engine that can undermine indexicality. Even simple retouching can do that, though it is the photographer making the choice. What the AI engine does is make the result so realistic that it does not appear to have been retouched, but I can do that in Photoshop, too. The Photoshop rubber stamp tool (and, for that matter, the healing brush) are standard tools for retouching photos, and both do actually also undermine the indexical relationships. But photography has always depended on the integrity of the photographer (or the processes surrounding the photo that preserve the chain of evidence) for its claim on truth.
I had better shut up now before Paul Raphaelson comes after me for getting it all wrong.
Rick "who spent a year reading Roland Barthes one week" Denney
