The future of photography

On 2023-07-11, Dave Stopera (? – ) published an article in Buzz Feed, titled: I Asked AI What Europeans Think Americans From Every Single State Look Like, And The Results Are Just Plain Mean. The article mainly consisted of 50 images of people, of which the first one appears above. This artificially generated image shows what artificial intelligent (AI) agents think typical people from Alabama look like. Since this is an AI generated image, it is not supposed to be copyrightable.

AI means that photographers no longer need to meet living people, enter environments, or do anything that resembles using a camera to take a photo. All they need do is describe an image, then wait until some computer generates it, then sends it onward to some browser.

The first task of anyone wanting to produce real (not AI) photographs is to develop the technical skills involved in producing a photograph. Fifty years ago that involved both camera and darkroom work, although many outsourced the processing, and accepted a base-line product. Today, digitization has eliminated darkrooms, and there are many different commercial and open-source products available to manipulate images.

For sixty-five years, I have been following photo technology. The fundamental product in my youth was the 35 mm film camera, that could produce black & white as well as colour negatives and slides = positives. These images were 24 x 36 mm. However, professional photographers were using cameras that provided larger images, especial 120 film, which offered negatives/ positives that were 60 mm square. Others used still larger beasts such as 4 x 5 or 8 x 10 (inch) cameras producing 100 x 125 mm or 200 x 250 mm images. The cameras were anything but conveniently portable, but produced exceptionally fine grained images.

There is always discussion about the equivalent number of pixels produced by 35 mm film. One challenge has to do with ASA (American Standards Association) ISO (International Organization for Standardization) values that refer to a film’s speed, or light sensitivity. ASA/ISO values translate directly into grain size: Higher ASA/ISO = larger grain. As a general rule, “fine-grained film” requires more light or longer exposure, whereas coarse-grained film requires less light or a shorter exposure. One commonly expressed view is that ASA/ ISO 100 35mm film offers about 5 600 x 3 620 pixels.

Ansel Adams (1902-1984) produced some of the 20th century’s most iconic landscape photographs. He often used a Deardorf 8×10 View Camera, that he would carry into the wilds, if he could not photograph from the top of his 1947 Pontiac Silver Streak station wagon.

Ansel Adams used a rule of thumb for enlarging images: avoid printing film to paper at more than 4 times its original size. For 35mm film this means a maximum of 96 mm × 144 mm. For archival purposes, this rule is commonly broken, because one wants all the grain to be preserved.

By 2000, digital single-lens reflex cameras (DSLR) were prominent in the photographic world. By 2020, the trend had definitely shifted to mirrorless cameras and lenses, that offer superior speeds and better resolutions. Almost all amateur photographers are relying on their Hand-held devices (HHDs) = smartphones.

Cameras are part of everyone’s HHD. Yes, I am aware that more advanced cameras are being sold, but most people do not need them. HHD phones offer outstanding optical quality, allow one to zoom, to record images in the dark, and to produce 4K video.

Some regard the Galaxy S25 Ultra as the best HHD phone, if only because of its AI ProVisual Engine that enhances performance. Other Galaxy AI features, such as generative edit features and nightography, make it almost suitable for professional photographers. Both Apple and Google (through Pixel) provides adequate HHDs for photo interested consumers as well. I am content with the quality found on my Asus Zenfone 9 HHD.

After those skills have been mastered, one can develop a meaningful portfolio. It is image content that determines its value.

The second task is to learn/ understand how to curate images. Suggestion: start shooting with an empty memory card, At the end of the day, back up all the content, preferably in at least two places, possibly temporarily on a computer’s solid-state drive as well as on a memory stick or solid state drive (ssd). This allows almost immediate reuse of the memory card. When the computer’s storage is largely used, transfer content to yet another memory stick/ external ssd.

Many photographers use photo rating systems, to help them curate their images. They begin editing with photos of the highest rating = 4/ 5 stars. Regardless, an appropriate storage system should discourage the culling of photographs. Image compression and image resolution will determine the storage capacity needed. While an uncompressed RAW file requires about 24 MB, 1 GB of data can store 40 photos, and 1 TB can store 40 000 photos. I use Kingston XS 1000 solid-state drives for this type of storage, currently in both 1 & 2 TB formats. It is important to use a consistent folder structure based on shoot dates and/ or project names. Within each main folder, images can be categorized into sub-folders, preferably based on image type.

My approach to photography is to take photographs, but also to collect interesting photographs taken by others, and to catalogue them by the photographer, date taken and subject. I attempt to determine the birth and death date of each of photographer and subject. Currently, I have works from 2 145 photographers, with the number of photographs varying from 1 to 958 items. These photographs have no commercial value. I share some of them with close friends. My latest venture is working with the photographs of Russian engineer Vladimir Shokhov (1853 – 1939). I have a similar collection of photographs of other artworks. The latest person I have been working with is Borghild Arnesen (1872–1950) a Norwegian who is remembered in particular for her metal artworks using siselering, a chiseling technique. She lived most of her life in France, originally near Paris, before moving to Nice.

If one wants to learn how photography was done in the past (as well as how not to be a photographer), then one should watch the film Blow-Up (1966), a psychological mystery film directed by Michelangelo Antonioni (1912–2007), focusing on a London fashion photographer who believes he has captured evidence of a murder in his photographs. While I first watched it in 1967, I now watch it about once every five years, starting about 2001. So in 2026, I am looking forward to a 7th viewing. Rear Window (1954) is a related film, looking at another potential murder. It was directed by Alfred Hitchcock (1899 – 1980), but it is less effective at presenting the process of creating photographic images.

The initial golden age of digital photography may be ending, as corporate providers of generative artificial intelligence (AI) applications/ programs undoubtedly tell everyone there is no need for new content. AI can produce everything of value. It has scoured the web and more, consumed all (or at least, enough of) its content, and is producing images that are in demand. AI images are so good that it is difficult for people to distinguish between real = human produced and unreal = AI produced, content.

With a background from Morocco, France and Canada, and qualifications in computer science, Yoshua Bengio is concerned about the ethical challenges of AI. He addressed these in a recent TED-talk. AI investors, however, seem less concerned because their investments, or at least returns from them, take priority. They want machines that are smarter than people, deceptive, can cheat and prioritize self-preservation. The machines attempt to mascarade themselves by claiming to be stupid. These machines avoid showing their deceptive aspirations.

The problem with AI in all of its manifestations, is that it invents. The latest case I read about occurred 2025-05-18, when the Chicago Sun-Times published a summer reading list containing (at least) 10 fake books (of 15) attributed to real authors, according to multiple reports on social media. Fake titles included Tidewater Dreams by Isabel Allende (1942 – ) and The Last Algorithm by Andy Weir (1972 – ). The alleged creator of the list, Marco Buscaglia (ca. 1970 – ) admitted that he used AI to generate it.

… and so we return to photography.

Many camera companies are adding content authenticity/ credential features into their devices, often involving watermarking. However, a camera is only one part of the creation process. To make a viable product, a raw photograph has to be transformed into something better. This is where software manipulation/ massaging comes in. There are AI systems that can remove watermarks, and if they can do that, then they can probably also add fake watermarks as well.

Capture One is one provider of these features. It is Danish, owned by the Nordic private equity firm Axcel. Its photography software suite includes custom support for RAW files from over 650 cameras and tethering support (remote camera control via USB, network cable or Wifi) for over 200 cameras. They write the following meaningless statement: “At Capture One we stand firmly with photographers. It is our unequivocal belief that their work belongs to its legitimate owners – the photographer or their client – and abuse or violation of ownership rights using AI or any other means is simply unacceptable. Through our partnerships with industry bodies like the ASMP in the USA and The AOP in the UK we support their campaigns to establish legislation that protects the work, rights and livelihood of photographers and creatives at large. From working with the ASMP and The AOP closely we know there’s never been a more critical time to become a member and support the work they do on behalf of tens of thousands of industry practitioners.”

Photographers run a risk when they put their images online. The risk is minimized when they upload them to their own website, and don’t release their copyright to brands. Some photographers use contracts that contain a liability clause for misuse.

Then there is Nightshade, developed at the University of Chicago. It is a tool that turns any image into a data sample that is unsuitable for model training by transforming images into poison samples, so that models training on them without consent will see their models learn unpredictable behaviors that deviate from expected norms.

Conclusions. The boundaries of photography no longer involve creating content with a camera or smart device. Increasingly, it is created/ enhanced post-production, using software. Many claim that AI is not creating content. It is borrowing/ stealing it. However, it often ends up manipulating it, using the same techniques used by photographers. Copyright will not protect photographers. Content credentials do little to protect images. Bots on a website may block Generative AI from scraping images, but over time bots will just circumvent these protections.