I have two passions in life - data science and photography. Both are modalities to enhance problem solving, decision making, and insight to ourselves and the world around us. Recently I have had really intriguing and frustrating conversations about the nature and boundaries of artificial intelligence.
As a compliance wonk, I fear that the regulatory limits of fair use and transformational work will be obliterated. As an HR professional, I am concerned about a widening earnings gap between technicians and creatives. As a photographer, I despair of the compromising of personal intent from artists’ work.
I recently tested several AI photo scaling tools and found them good but not great. Two softwares changed the shape of my client’s eyes. Several changed the light points in the irises. Most sharpened beyond reality and pushed the image into the uncanny valley. Photo enhancing has always existed; anyone who has dodged and burned in a darkroom has done it. But AI tools take the decision making somewhat out of the creator’s hands to create a new reality.
In college I asked my photography professor “what is the difference between a photograph and a snapshot”. We discussed at length and determined that the key difference is intent; a snapshot is to record, whereas fine art photography is to express. It is a reflection of the artist; of their relationship to the environment and the subject. My father is a photographer and we often review each other’s work. He loves a photo I took on vacation which he calls Cirque Du Vegas. He is drawn to the colors, the lines, the juxtaposition between architecture and nature. I find it hard to appreciate as an art piece because I know the intent behind it - I was testing a new memory card. The image does not speak to me on an emotional level because the personal context is analytical. Which begs, should this then mitigate a viewer’s appreciation of the work on its own merit?
These conversations are coming to a head and will soon play out in the UK and Delaware court systems. Getty Images has filed a lawsuit against Stability AI for copyright infringement. AI art generators do not create images from whole cloth, they do not draw on blank canvas as an illustrator would. Behind the scenes are many millions of images fed into a model without permission or compensation.
From a workforce perspective, what will the impact be on the creative economy? A sticking point in the Writers Guild of America strike is the use of large language models to replace human tv and film writers. The Recording Academy issued new guidelines about Grammy eligibility for music with AI elements. The 2023 Sony World Photography Award granted an AI generated image first place. (note: the artist Boris Eldagsen declined the award and stated that his entry was to “spark a discussion”). New AI headshot services promise to “transform your selfies” for $30, a fraction of the $400-$2000 skilled photographers charge for a portrait session.
As cameras improve on smartphones and digital imagery becomes the preferred lexicon, societal appreciation for photography as both art and science has diminished. As a photographer my copyrights are violated constantly on large and small scales. Clients proudly stated that they printed low dpi and watermarked images on plain letter paper to put on the fridge rather than paying the reasonable 99 cents for a print. Gallery curators digitally altered my images for advertising without credit. Website visitors took screenshots to use as lock screen backgrounds.
At the intersection of art and technology lies the question - just because AI can, should it? How do we explain the human costs to the scientists creating these models? An artist’s heart is unmeasurable and talent is invaluable.
Ultimately the question is how do we utilize machine learning and AI to honor and enhance the integrity and humanity of artistic expression? The ability to translate a vision into an effective prompt is laudable, but does not make the prompter an artist. Herein lies the boundary between skill vs talent.
Please share your thoughts in the comments and join me next week for proposed recommendations.