Examining historical timelines is always a captivating exercise, particularly in our modern era, enriched by the revolutionary invention of the World Wide Web by Tim Berners-Lee. It prompts us to reflect on how those who came before us might perceive today’s technological advancements. Consider William Henry Fox-Talbot (1800–1877), who once remarked, “I do not claim to have perfected an art but to have commenced one, the limits of which it is not possible at present exactly to ascertain.” Would he have regarded the emergence of ChatGPT in 2023 as the ultimate boundary of his art, or simply as another milestone in its ongoing evolution?

Photography’s Embrace of Change

As photographers, we have witnessed and experienced countless innovations throughout the years, often meeting them with a mixture of scepticism and curiosity before ultimately welcoming the changes they bring. With the arrival of Artificial Intelligence (AI) in the creative sphere, there were, perhaps inevitably, predictions that photography’s days were numbered. Yet, history reminds us that each new development is less an end than a new chapter in an ever-unfolding story.

Introducing AI to Stroud Camera Club

With an eye towards embracing technological advancement, in late 2024 I conceived a novel approach to integrating Artificial Intelligence within Stroud Camera Club. Founded in May 1939, the club’s origins trace back to a group of pioneering enthusiasts who established its presence just as the world stood on the brink of World War II. This rich history has always inspired a spirit of innovation among our members.

Reimagining the Club Challenge

In recent years, our club nights have featured a popular three-way team challenge. Participants are randomly assigned into three teams, with each photographer contributing at least one image to represent their group. Judging is typically carried out by all members using a paper voting system, with scores meticulously compiled in a complex spreadsheet during the break to determine the victorious team.

AI Takes the Judge’s Chair

Seeking a legitimate way to introduce AI, I proposed using ChatGPT—affectionately dubbed “Chatty” by club members—as our judge for the challenge in 2024. This experiment proved to be a resounding success, garnering enthusiastic feedback from participants. The positive reception led me to share the experience on the UK Photography Facebook page, sparking interest from several other clubs.

A New Challenge Emerges

Our innovative use of AI caught the attention of Cardiff Camera Club, who reached out to propose an inter-club “Battle” judged by AI. They suggested hosting the event on Thursday 12th February 2026, the regular club night for both organisations. True to our adventurous spirit, Stroud Camera Club eagerly accepted the challenge, looking forward to this new chapter in our ongoing story of adaptation and change.

Building the Foundation for AI Judging

The challenge then became apparent: How can I effectively structure an entire evening, ensure that both clubs view all images through large-screen projection for optimal experience, preserve the social element that is often diminished during remote sessions, collect submissions, obtain AI-generated critiques and scores, and have these results announced simultaneously to both clubs?

The planning started.

At Stroud Camera Club, we have been using the online competition system Pixoroo. This was set up by Cheltenham Camera Club member, Aleks Gjika, “by a photographer for photographers”. It is web-based and we have worked with Aleks on various enhancements. Being an IT professional, Aleks was keen to include the option of an AI judge within Pixoroo. At this stage it is simply the functionality of the judge not being real, and the facility of a text to speech interpreter within the software. This gave me a starting point on which to build the event.

My AI of choice for photography critique is ChatGPT – there are plenty of other AI models, and I do periodically use Claude AI, Gemini, CoPilot to name a few – but I generally find that ChatGPT has a better understanding of the camera club world, and tends to read an image in an authentic rather than purely generic manner.

I knew however that one of the most important aspects of utilising AI to its fullest is the prompt. ChatGPT offers a “Project” module, so that was perfect for starting to build a suitable prompt. I spent several hours (even days!) working with sample images from our club’s archive to enable ChatGPT to understand what I was attempting to achieve. Ideally I wanted all the images to be viewed first before any scores, but I was happy for the critique to be done as images were uploaded. I needed to explain within my prompt that I needed fairness and consistency, I also wanted constructive critique and encouragement. Precise wording is necessary even to explain that at camera club we do not generally give scores of 16.75!

The eventual prompt was this:

You are a UK camera club judge.

The competition is styled as a battle between two camera clubs.

The judge will offer around a 90 second conversational critique of each image and give a score between 10 and 20. Half points are not available. The scale of scoring is technically 0 to 20.

The judging should be consistent, fair, constructive and honest. The judging should consider all aspects of camera club photography, so to include technical, composition, emotional intent, artistic statement etc. No specific bias should be given to any of these aspects, the images should be interpreted individually and only judged against one another at the point of scoring. That having been said, the images that are likely to score lower should not be given a commentary that sounds as if they are going to receive higher scores. The judge should remember that this is regional camera club level and not International level - although no comment about this is required. The judging should not give any inherent bias to one genre eg. nature images do not score higher or lower than any other genre.

A few of the best images should be given a 20.

Each image will be uploaded individually for all the critique first and then when all the critique is complete you should give the scores for consistency and so you can assign a couple of 20's and 19's for the very best images on the night.

The battle is between 2 clubs each will be submitting 20 digital images. The images are uploaded in randomised order and with no identification of the club nor photographer.

So, leave this bit to ChatGPT and on with the plans!

Managing Competition Entries with Pixoroo

Obtaining and Organising Club Entries

Acquiring all the competition entries was a straightforward process thanks to Pixoroo, an online competition platform. Each club was able to upload their submissions at their own convenience, right up until the closing date. Pixoroo’s system is particularly suited for a Battle format, offering a dedicated module with clear, intuitive prompts and automated email notifications for all participating clubs.

Image Submission and Download

Upon completion of the submission window, each club had successfully uploaded their allotted twenty images. At the close of entries, all images were easily downloaded from Pixoroo to my computer with a single click, adhering to the competition’s required resolution of 1600 x 1200 pixels. Each image was subsequently uploaded into ChatGPT, which I was able to use on a free month’s trial—ordinarily, the service is available at a cost of £20 per month with the option to cancel at any time without further obligation.

Critique Process and Output

ChatGPT provided critiques for each image in accordance with the carefully crafted prompt. The platform features a convenient “copy” function, allowing the critique text to be easily transferred and pasted directly into the appropriate field within Pixoroo.

Alternatives to Pixoroo

For those not utilising Pixoroo, the critique text can simply be retained within ChatGPT itself. The platform offers a text-to-speech module, which means that for internal competitions, the feedback can be played directly to an audience. This was the method used for our initial internal event. Alternatively, a person could read the text aloud, though this might somewhat diminish the sense of automation and impartiality.

Image Randomisation and Scoring Prompt

To introduce an element of randomness when using ChatGPT, images were uploaded based on their file size. Alternatively, there is the option to use Pixoroo’s built-in randomisation feature, which allows images to be uploaded according to those parameters instead.

At the conclusion of the critique phase, the following prompt was provided to ChatGPT:

So you have now seen and critiqued all 40 images. You now have to fairly and consistently allocate the marks. Remember this is a Battle, so these are some of the top images from the two competing clubs. You can (and should) award at least one score of 20 and indeed if warranted you can award multiple images a score of 20. The scores do not have to be spread evenly across the range - but they do need to take into account which are the better and more competent images that offer the viewer the fullest experience, compared with the less competent images that are lacking in quality, emotion or other aspects. Please give me the scores in a list with the title beside the score. At this point I do not need justification, just plain scores.

Running the Competition Night

Technical Setup and Club Participation

With all images, critiques, and scores prepared, the next challenge was to organise the event in a way that preserved the clubs' social atmosphere while allowing simultaneous broadcasting of images and critiques. Following careful trials, the chosen solution was to set up a Teams meeting (with Zoom as a viable alternative) hosted on a single laptop dedicated to running the competition. Both clubs participated by joining the meeting from their usual meeting rooms, with their own laptops connected to their respective projectors and sound systems.

Optimising the Workflow

This approach ensured that no single laptop was overloaded with multiple tasks, thereby maintaining stability throughout the event. The key was to keep both sound and microphone muted on the host laptop, especially as it was positioned in the same room as the Stroud Camera Club’s projector laptop. Pixoroo operates on an “extended screen” basis, displaying images and scores on the projected screen, while the operator manages the competition controls separately. By dragging the relevant window into a separate browser window and sharing only that window on Teams (with sound enabled), it was possible to effectively replicate the extended screen experience. There were a few brief, busy moments when Cardiff joined the meeting and needed to adjust their sound system, but once resolved, the event proceeded smoothly.

Evening Proceedings and Feedback

The evening itself ran flawlessly. Pixoroo automatically converted the written critiques to voice, which were then played over each image for both clubs. As with any judging process, there were some differences of opinion regarding the critiques, but the majority of members responded positively, noting that the AI judge highlighted many areas for improvement. The platform’s timing display showed precisely how long each sound clip would last, ensuring that each image received a similar length of critique and commentary—a consistency not always achieved with human judging.

Scoring and Results

The scoring phase was conducted with consistency and fairness, culminating in a competition that remained tightly contested until the very last image was shown. Throughout the event, Pixoroo efficiently managed the entire scoring process, from calculating totals to generating individual results sheets for each club. Its reliability and accuracy were evident, affirming its usefulness as a central tool in the competition.

Notably, four images achieved the maximum score of 20, with both clubs each contributing two of these top-scoring entries. This outcome highlights the strong competitive spirit and quality of submissions from both sides. Importantly, the evaluation was carried out by ChatGPT without knowledge of the image authors or their club affiliations, demonstrating strong impartiality in the judging process.

Reflections on the Role of AI Judging

It is clear that an AI judge will never fully replace a human judge. The absence of emotional intelligence and the nuanced understanding that a human brings to the judging process cannot be replicated by artificial intelligence. However, the evening demonstrated that the inclusion of an AI judge contributed greatly to the enjoyment and engagement of the event. The session offered both clubs an opportunity to embrace modern technology in a fun and entertaining way.

This development holds particular significance at a time when many clubs face challenges in attracting younger members, with some clubs unfortunately closing their doors. It serves as a pertinent reminder of the importance of staying up to date and embracing current technologies and practices. While it is vital to acknowledge and respect the club’s rich heritage and the pivotal events of the past—such as the groundbreaking developments in photography in 1835—it is equally important to look ahead. Only by balancing tradition with innovation can clubs continue to thrive as vibrant, friendly, educational, and enjoyable spaces for all members.