If you were to ask 10 photographers about their opinions of artificial intelligence in photography, you just might get ten answers.

With that, we might get different emotions. Excitement, anger, betrayal, fear, anxiousness, indifference. Remarkably, all these reactions might be valid. And these viewpoints might morph over time. I’ll go over some thoughts of mine as we enter the infancy of AI in photography.

Alleviating tedious post-processing

Currently, what interests me most about artificial intelligence is greatly reducing repetitive, time-consuming post-processing. I love nailing the photo in-camera as much as possible. However, just like a negative, RAW files need to be processed. And many RAW files can take many hours to be processed.

What if?

  • What if detail-oriented, painstaking processing or batch-processing could be done effectively by AI?
  • What if a sports photographer or wedding photographer with giant time constraints could batch process their photos? Perhaps AI could learn their processing preferences and batch process hundreds of photos?
  • What if a portrait photographer could get, say, 80% of the way to where they wanted to go by using AI to batch process headshots? Imagine asking it to choose all the photos where the eyes are in focus. Then with only those, eliminate blemishes, flyaway hair, bloodshot eyes or other parameters?
  • What if AI could effectively create luminosity masks that clearly delineate between items, knowing what to mask by us giving commands such as “mask the background” or “mask the sky.”
  • What if AI could learn who specific people were quickly and automatically find all of those images for you instead of tedious keywording? What if AI could input those keywords for you?

  • For night photographers like me, what if AI could learn what the sky was and apply noise reduction without affecting other parts of the image? What if AI could differentiate between hot pixels and stars and eliminate the former? What if it could create masks for me by determining what I was looking to mask?

We all might be able to spend more time doing other things, such as more photography, marketing or <gasp> spending time with our families and friends.

AI is already in our mobile devices

AI and machine learning has already found its way into mobile phone processing. Portrait mode on iPhones, for instance, use machine learning and AI to determine where the background is and then applies a blur, or bokeh.

The traditional method was to use a lens with a wide aperture. If done effectively, do you feel it matters whether it’s done with a lens or with AI? Are these simply different methods that accomplish the same thing, or is there something inherently wrong with using AI for this?

Blurring the creative boundaries?

The promise of AI is for creativity as well. The minds — the human ones — behind software such as LuminarAI, Adobe Sensei and ON1 promise to jump-start creativity in a big way.

LuminarAI promises to offer suggestions on cropping. This was always in the hands of the person doing the post-processing before. But if AI offers suggestions that follow the Rule of Thirds or the Golden Ratio to help reduce post-processing time, is this good?

Sky replacement

Luminar 4, ON1 and Photoshop already offer sky replacement. These sky replacement options can also relight the foreground, making the blend more realistic.

Some landscape or real estate photographers might welcome this. Imagine that you are on a vacation, and every day has horribly overcast skies. Or a real estate photographer has to make a house look great, but there’s nothing but horrible hazy skies. Would you welcome the chance to transform your ho-hum photo into something more attractive?

I’ve used sky replacement before for my night photos. But I used it for a different reason. When photographing Milky Ways, I create low ISO photos of the foreground to reduce noise. And I sometimes “stack” my Milky Way photos, also to reduce noise. Then I blend them together. Luminar 4 to the rescue!

I have used their Sky Replacement tool to drop in my own sky tedium of creating a mask myself. Although Luminar 4 has great difficulty discerning a dark sky, I’ve managed to do it once or twice. In this case, I am using its Sky Replacement feature to work more efficiently.

But what if everyone starts replacing their skies just because, well, they want a different sky? Is that good? Is it photography? Is it art?

AI as dominant creative force?

As I mentioned, AI is in its infancy. But we have already seen the startlingly real “deepfake” videos and how, increasingly, it’s difficult to tell reality from fiction. How will this impact social media? How will this impact a public that already has struggles with discerning “fake news” from scientific fact?

AI is already making creative decisions. And although we like to say, “No worries, it’s just another tool,” is this really “just another tool?”

Artificial intelligence may be more than another tool. When AI creates things, offers creative suggestions, and more, this becomes more than “just another tool.” AI becomes a creative force in itself. This may be unnerving for many of us who grew up thinking creativity was the sole domain of humans. And it may be unnerving if it eats away at our paychecks as well.

If AI can learn the tendencies of a creative individual and create templates, suggestions, looks, processes, actions and more in software, creativity is no longer the sole domain of humans, but also of machines.

What if AI begins creating novel images after someone inputs certain parameters? Should we regard an AI-produced image as art if it has taken numerous images and created its own original composite, lighting it and all? What would this do to jobs in photography and post-processing? Advertisers and many in commercial industries might embrace this royalty-free way of working.

How do you feel about AI?

What concerns do you have about AI? Are you excited about the possibility of working more efficiently? Do you feel it’s good, bad, a little bit of both?

Will it eventually rob photographers and post-processors of jobs? Do we want AI eventually creating completely novel images? Leave your thoughts in the comments below!