AI and Human Portraiture
Why Artificial Intelligence will never replace human one-on-one portraits.
AI will never replace human portraiture. I’m going to explain why by sharing my recent experience using Midjourney and ChatGPT to create images as a busy art director.
When ChatGPT 3 (and then 4) were made available to the public, I was working as director of a publisher’s media department. I had built the department from the ground up, and was tasked with overseeing three projects: manage and grow our neglected flagship blog; produce 35 videos a year, both shorts and mini-docs; and launch 6 podcasts, including a flagship pod that I would host, manage, and produce. I was buried by the amount of work because there was little to no budget to hire outside help, and I was quickly burning out. Then came AI.
I immediately saw the benefit of ChatGPT as a leveraging tool for my responsibilities with the blog. I learned how to effectively prompt and draft blog-friendly and SEO-optimized articles for our niche customer base. Amazingly, after some editorial oversight and a bit of research, many of these articles made it through our theological editor’s desk with minor corrections, and were subsequently published. I knew I’d passed the test when the response was, “Great article, who wrote this?”
Then I turned my attention to the image generating capabilities of AI. I needed header images for the blog, because most stock photography is lame. The learning curve was quick, and my background as a fashion photographer enabled me to create in-depth prompts that paid attention to lighting, focal lengths, even camera and lens profiles that resulted in some stunning imagery. Here’s an example, a photography taken on a Hasselblad camera of a the Beast from the Book of Revelation:
And another one, this time using a Fujifilm GFX100 camera and a 32-64mm lens (the same kit setup I employed to shoot fashion in London):
Since leaving that role, I’ve gone back to shooting actual photographs of real people. Although I do offer elite-level AI image generation as part of my production company’s services, all of my portrait clients want pictures of themselves, the actual person, and not an AI-generated avatar.
The return to real-person portraits from AI-generated images jolted me in a way I hadn’t anticipated. Whatever pleasure I derived from creating fantastical, realistic imagery using AI is overshadowed by the joy of capturing real human people in real life moments.
In fact, there is nothing in my life as an artist that comes close to the experience of portrait photography due to one singular phenomenon: intimacy. This precious thing, intimacy, is only possible between two (or more) humans, especially within the context of a mutual creative act.
On the set of a portrait shoot, whether it’s held in somebody’s living room, outdoors on a mountaintop, or in a professional studio, the people working together in that space and time enter into a dance of sorts.
The photographer is guiding the model, encouraging her, leading her into different bodily and emotional expressions, positioning her in the proper light, and this is done through use of hands, words, eyes, and movement together.
The model is also engaging with the photographer, responding to his guidance, moving where he places her, opening herself up emotionally and physically; away from inhibition (freezing up) and towards authenticity of movement and expression.
These phenomena cannot be replicated in a Large Language Model. The subject’s eyes in a visual representation created by AI will never achieve humanness, only merely a mimicry of it.
Because although at a fundamental level we are simply talking about pixels, those little dots of light or ink that make up an image on our screens or printer paper, the intimate dance of human-to-human portrait photography gives to the resulting image a glimpse into the imago dei, the image of God impressed on every soul.
And the miracle of a portrait is that, although only one person is visible in the final frame, it has been created by the intimate engagement of two people who both carry that same imago dei.
That is something of which even AGI will never be capable in realis, only ever in simulacris.
*Stay tuned for my next post expounding on this last point about reality vs. imitation, and the implications for so-called “truthful AI” (and why its antithesis, deceitful AI, is closer to the core of AI entirely).