Many cameras and software products these days are claiming artificial intelligence (AI) is built in. Unfortunately, that term doesn't mean much on its own. Most companies are using "AI" as a marketing term, not as a precise description of some technology that's being done behind the scenes.
Artificial Intelligence is really a catch-all term for a lot of very different data analysis and transformation approaches. Many actually use something more precisely called machine learning (ML), which is a process by which the "AI engine" is trained (programmed) using many passes on a set of different inputs. Edge or object recognition is often done via ML, for instance, because you can tell the system "that's a cat, that's not a cat, that's a cat, that's a dog which is not a cat," and so on. With the right number of passes (too many actually tends to lower the results), the ML system begins to be able to do this on its own with some high (but not perfect) degree of accuracy.
But that's not the topic of today's article. What triggered me was an example used for a new program, which claimed to create useful compositions from your image data. One example they showed had two vertical elements within the scene, and, you guessed it, it created a vertical composition.
Now I happened to start out mostly as an editorial photographer. I developed this "AI" in my brain because my photo editor hit me over the head when I didn't give them the same basic photo as a vertical, as a horizontal, and as a vertical with a free space area at the top for the magazine's logo. Adding detail shots was the next challenge from said photo editor.
"If you're not bringing me back at least five or six different choices for use in different ways in the magazine, I'll find another photographer," he said to me.
So I'd wander around looking for photos for the magazine, find something that I felt needed capturing, and then do essentially what he said: find the vertical in the scene and shoot it for both a full page use and a cover use, find the horizontal in the scene for use as a double-truck (two-page spread) or a half page opener, and find the details within the shot that were important.
I call this Intelligence. Or maybe Human Learning ;~).
Okay, now I can get to my point: many of you are simply wanting automatic functions to do the work you already should be doing. The reason for that is that you're overwhelmed by the process of taking and presenting a photo. You'll take "close enough" or "good enough" because you can't get to a higher level on your own.
This is really what separates the smartphone crowd from the dedicated camera crowd, or should. Dedicated cameras can (and should do more to) open up all the possibilities for an image that we're capable of capturing. Unfortunately, the camera companies keep getting caught up by the fact that more people will buy the "good enough" automatic functions than the advanced functions that you, the photographer, needs to control precisely in order to get exactly the "right" image.
I've written this before, but it's worth repeating here because it illustrates the issue: if you don't want to take the time to fully master a high-end autofocus system, the current Sony Alpha cameras probably do the best job of guessing what it is you want and getting "close enough" results. You'll see people claim that the Sony cameras get perfectly focused results in bursts on moving objects. They don't. They get "close enough for the person writing that" results. Under control of someone who really knows what they're doing and moves between focus choices based upon training, the Nikon D6 is the best focusing camera I've encountered. The top Canon's (1DX Mark III, R5, R6) come very close. The Sony A9 Mark II is a visible step behind, and some of the other Sony cameras a step beyond that.
Anyone that's followed my writings on the Internet about photography for the past 25+ years—yes, it's been that long—knows that I always seek out best. Not just best in equipment, but best in me, as well. I write about optimal data capture and optimal data processing, basically. The pandemic has been a real problem for me because the types of things I need to do to stay my best in the field are harder (or in some cases impossible) to practice, let alone perform for money. So as the pandemic wanes, I'll have some real work to do in order to get back into top shooting shape.
And that's going to involve improving my Intelligence (or again, Human Learning). If you don't practice things, then you don't develop the muscle/brain memory necessary to keep those skills polished.
I'm not against AI, ML, or automated features. At times they have their uses, and for some they produce better than they'd otherwise get. But marketing these things is a bit like putting you on a Holy Grail quest. What you seek, you might not find. Here's another (strained) metaphor: there's no free gourmet lunch, but sometimes you can find an inexpensive and "okay" meal that suffices for now.
