You can’t always get what you want

Futuristic AI Robot Spilling Coffee on White Kitchen Wall and Co

Artificial Intelligence of course is the buzz word of this quarter of the 21st century to the extent that you can’t seem  to launch any product without attaching it as a prefix or suffix at the risk of appearing outdated if not completely redundant.

It has always been so. Once upon a time the name digital promised progress and modernity long before it became tainted as something that couldn’t be repaired or updated when the software passed its sell by date. 

Any advance in technology has equal measures of benefit and cost -often only realised in the long rather than the short term. At the moment opinions seem to be equally divided.

Personally my position is similarly both positive and negative. As a photographer I took on the challenge of digital without rejecting the appreciation of traditional film. That progress led me to get more involved with printing than simply dabbling in the darkroom. From desktop to large format the inkjet process created a much closer understanding of how electronic imaging translated colour and detail, giving me not only a second career in direct print, but a much greater knowledge of how a modern camera interpreted the world.

And that last effect is important as my relationship with film - despite 25 years working with it - has never been to consider it magical in itself as many still do. It was always an optical mirage turned into reality by a chemical reaction. Any magic was in the eye of the individual photographer to use that stored potential into something special, and often something quite unique not seen immediately by the naked eye.

Back in the days of film proper photographers were few and far between. Even enthusiastic amateurs would only shoot a few dozen rolls a year - hardly a couple of hundred images in total. That was the practical and affordable limit of the technology of the day. This meant a steep learning curve to reach professional or creative standards simply because it needed time behind the viewfinder to achieve results beyond a lucky one-off.

The early days of digital were perplexing as no one had any previous experience, and although so many more people were able to capture images, the shared experiences only seemed to confirm that the process was much more random and unpredictable. It was convenient, but chaotic in its output. This is why some simply condemned it as a novelty, and others new to the system simply gave up as it couldn’t reproduce the magic wand effect that had been promised.

This was all very unfair as in comparison film had had over 150 years of development and I’m sure the  early Victorian experiments suffered just as many frustrating failures trying to make sense of the mysterious method of image transfer. The major difference is that once the process had been more or less understood, and the materials perfected, any improvements were mostly incremental. Film could last for years long after its sell by date and mechanical cameras, like my 40 year old Nikon, were still perfectly serviceable, with or without batteries.

Digital devices, on the other hand, and the software that goes with them date very rapidly as we have learned to our cost. That is in the nature of the technology which expands exponentially - that is in every direction all at the same time like some expanding universe. AI in this context is just another big bang along the way, not the final solution. It’s a process, not a product, and love it or loathe it that process will progress relentlessly in the background of many of the things we work with. 

That is why it is important to have at least a basic understanding of how it works to understand or possibly anticipate some of the unexpected consequences that may arise. AI is not just a faster computer programme, it’s a smarter way of exploiting a much greater amount of digital information than previously possible, to achieve unprecedented results. But the unexpected element is itself the possible flaw in the process. Almost everyone wants things faster, and better. But be careful what you wish for, as they say.

As a long term Photoshop user I have, in some small way, been helping Adobe develop its own AI over the last decade so I guess I’m partly responsible for wanting tools that transformed images so much better than the painful manual operations of years gone by. Compared to some of the early versions, the modern PS control panel is as different from that of a 1920s Model T Ford to a modern Formula One car. It still steers the vehicle, but the computing calculations, and the possibilities, are beyond comparison.

I often use the example of driving as it’s something that demands a significant amount of essential human skills - direction, anticipation and control - over a machine. Even those promoting driverless vehicles, with computer sensors in control of these actions, accept that there may be times of crisis or uncertainty when a human has to intervene. But that is precisely the reason for caution over an entirely autonomous process as how is a person, increasingly detached from learning practical solutions, prepared for the unexpected ?

Serendipity is that discovery of something valuable by chance rather than expectation, and is part of very human learning and the accumulation of experience. A chance discovery remembered  may later trigger a lightbulb event to provide a solution. Our brain cells are capable of thinking outside of the box. I’m not sure whether AI, or at least the way it is being employed, is quite as flexible. Hopefully I’ll be proved wrong but it does seem that many of those championing it as a one-stop solution have limited experience of using it, or have not explored possible alternatives.