AI In Photography, The Good, The Bad And The Ugly

It feels like the last couple of years the words artificial intelligence have dominated the photographic world. Camera companies, editing tools and stock agencies have rushed to embrace the AI revolution, often with a seemingly blatant disregard for the views of real world photographers.

However, whether we photographers love or loathe the rise of AI, there is no going back. That horse has bolted and we now have found ourselves in a position of how, when and why to embrace AI in our photography.

Not all AI is a force for bad in the photographic industry so today we are going to look at the good, the bad and the ugly. In order to end this article on a more positive note, I will start with the ugly.

The Ugly Aspects Of AI In Photography.

The obvious choice would be AI imaging, however I am going to put that into the bad, not the ugly. The real and most potentially damaging aspect of AI imaging is misrepresentation.

This can range from relatively harmless social media engagement farming all the way to the manipulation of politics using deep fakes. The latter can be blatantly obvious or it can be very subtle. An example of subtle manipulation is the use of deep faked old images of a country suggesting how things were so much better back in those days.

On Facebook the use of AI imagery to farm social media engagement has become rampant. Typical examples will be very obviously fake AI images of, for example, London in the snow. The images look beautiful to someone that knows little about London but to anyone who has ever been there, or to any photographer, they are blatantly AI. However there are huge numbers of people that believe these images and like, comment and then follow these pages.

All very innocent you might think, after all following a page of pretty pictures, is not harmful. However, the intent of these pages is often not that simple. Behind them are often companies or political groups that once they have enough followers, will change the name and tone of the page to suit their needs.

There are also deeper problems with the use of AI within the photographic community.

First and foremost is the homogenization of images. With stock agencies now licensing AI work, we are in danger of a world where websites and adverts are full of vaguely similar, faintly realistic images that actually have no relevance to real life. Of course the driving factor for this is cost, but by keeping the spending low, designers and advertisers risk losing the creativity that the photographic world brings to them.

The Bad

Whilst the use of AI images in social media and political campaigns is ugly, the general use of AI in photography I feel is bad but not surprising. The worst aspect of this is photographers, or indeed even non photographers passing off AI images as real photographs. Landscapes and wildlife seem to be particular favorites as they can appear a little more authentic than images of cities and towns.

Another concern is the potential loss of skill sets. Whilst enthusiast photography will continue, the demand for commercial, professional photography is at risk and with it the skills, talent and creativity of a large number of professional photographers.

There is also the issue of the use of AI within real imaging. Personally I differentiate between the use of AI to add elements that didn’t exist and the use of AI to aid editing. The former is bad, the latter is ok.

However, increasingly, some photographers are using AI to add elements to real photos that did not actually exist. One of the more obvious is the addition of light beams to twilight shots of lighthouses. However there are plenty more examples. Whilst I am not against this, I do feel photographers should be upfront about the use of AI to add to a shot.

There are also very real issues around copyright within AI imaging. Most non-photographers believe that an AI generated image is created entirely inside the CPUs of a vast computer. However, that’s not the case. AI has to be trained, and the only way it can be trained is by looking at real photographs by real photographers.

A significant number of the main AI companies have trawled the Internet, harvesting images without permission or recompense, throwing up also sorts of legal complications. It’s quite possible that you might see an AI image that looks startlingly similar to one of your own and not have any control over how it is used.

There Is Some Good News.

At its simplest, we can use something like ChatGPT to inspire us when we are in the photographic doldrums. Simply tell AI what types of photography you like doing, and ask it to give you suggestions. This can be for a simple one hour shoot or for a year long project.

AI chat can also give us good answers to technical and even creative questions that we have about photography. Because photography is such a well trodden path online, the answers are often surprisingly comprehensive and accurate.

We will increasingly see the use of AI built into our cameras. My Sony a7Rv for example uses AI for focusing. It can determine the difference between multiple different subject types and predict their movement. AI will be incorporated into metering, white balance and quite possibly in the future, as a compositional aid.

AI is increasingly being used in editing software. One of the most powerful tools I use is Lightroom’s AI denoise. I can now take 61mp images at 12800 ISO, run them through the denoise and get almost perfect, noiseless photos.

AI removal tools take a lot of the legwork out of cloning out blemishes or unwanted elements. We can automate tasks more easily, especially useful for photographers having to work on large batches of images. Selection tools have also become much easier to use with the advent of AI.

The use of AI in photography is a tricky and controversial subject. However, it is not going away and will only increase over time. As photographers it is our duty to define the way AI evolves, to be part of it rather than a victim to it. That is very much the challenge ahead for all of us.

Further Reading