Yesterday, I included this photo of three happy puppies. If you saw it, you may have had a moment of feeling a little creeped-out by it:
There’s something slightly wrong with these puppies. Their eyes are oddly glowing, yet lifeless. Their tongues look more like slabs of bubble-gum than doggy tongues. Something about the combination of focus and blur, the slight asymmetry, the indeterminate edges of their faces...
It’s unsettling.
Why?
Because AI.
I asked Craiyon, a free online AI image generator, to create a photo of “three happy puppies.” And it did. Sort of.
I needed a picture of three puppies, and I didn’t have one. So AI made one for me. But I wouldn’t recommend using this image in real life. Because it’s creepy.
But you and I both know that anything digital is going to get better. And quickly.
AI-generated images are already less creepy and more realistic than they were last time I looked into them just a few months ago. Back then they were surreal and hilarious. Now they’re just sort of uncanny. Eventually: hard to tell from real photos.
But there’s something else to consider. You might be tempted to have AI create an image of people ... because you don’t have the image you need. Or the need to respect real people’s privacy makes real photos impossible.
So you could instruct AI to give you a photo-like image of Rohingya refugees in camp, mother and baby, smiling. You could be as specific as you need to be. And you’d get something.
But not a real person. A totally made-up image. It may have the elements you need to give your audience the human experience you want them to have. But it wouldn’t be with actual humans.
Would that be ethical?
I don’t know. But it feels borderline at best. It seems at the very least, AI-generated images should be identified as such.
Think about it, because it’s likely to be an issue for you eventually.
Recent Comments