As much as the pirate ship in Elliott Bay excited me, it was the fuzzy little bees.
Samsung wants us (and its shareholders) to know that its new phone is its most AI-powered phone yet. The Fold 6 I’ve been testing has a new tool called “sketch to image.” Draw a rough sketch on a photo or a blank note page, and it’ll use generative AI to turn that sketch into an image. When Samsung announced this onstage at Unpacked, I dismissed it as just another AI thing, but guys, this is seriously good. So good that it’s a little worrying.
Using the sketch-to-image tool in Notes is no problem. Draw something, highlight it, and then choose from a few styles, like “3D cartoon” or “illustration,” to turn your doodle into something more detailed. The image is sent to the cloud, and after a moment, you’ll see a few options. The results are usually cute and fun. I’ve drawn goofy dump trucks and school buses in response to requests from my 2-year-old. Sometimes I’ll draw a teddy bear with too many arms, but nothing too serious.
Something strange happens when you paste a sketch into a photo as an image. I’m the worst artist in the world, but this tool transformed my basic sketch into a photorealistic image, with AI-generated elements realistically integrated into the photo, sized, and adjusted to fit the surroundings, making it hard to tell they’re fake.
Which brings me to the bee problem. I was on a pier just south of downtown Seattle and I took a photo with some flowers in the foreground. The flowers were close to the camera and the focus was far away, so they’re a bit blurry. I drew the world’s worst bee sketch on one of the flowers, thinking the AI would insert an image of a bee in focus that would give it away as a fake. I was wrong.
The AI bee is just as blurry as the flower it rests on. If you didn’t know about the AI bee’s origins, you wouldn’t think twice when scrolling through its images on Instagram. You’d assume the photographer took the photo at just the right time, or was waiting for the bee to fly into the frame. This is something that takes skill and patience. But it’s not. In fact, we’re not even sure you’d notice the “AI-generated content” watermark in the corner of the image.
At first glance it looks convincing, but look at it for more than a second and you’ll notice something’s wrong.
I’ve been playing around with Sketch to Image for the past week, and the results aren’t always “fuzzy” or nice. They often have the telltale signs of generative AI art – words scrawled in an alien-like language, weird textures that just don’t look right, etc. They look convincing at first glance, but if you look at them for more than a second you realize something’s not quite right.
Sometimes the content itself reveals the truth — I don’t think anyone would believe I saw a giant pirate ship parked in Elliott Bay, or a giant orange cat at a West Seattle intersection — but even when the image is so outlandish that no one would mistake it for the real thing, it still looks real.
Generally, anything big will look obviously fake. But it’s very easy to add another car to a photo of a busy road or a yacht in the distance, and most people won’t notice it. Aside from the AI watermark, which can be easily cropped out, there’s really no way to know if there’s anything unusual in the image. Weird!
An unfocused bee is not going to destroy the fabric of our society.
I don’t want to get too carried away. The sketch-to-image conversion is completely optional, and many people can’t even find it in their Gallery app. Out-of-focus bees aren’t going to tear down the fabric of our society. But I think AI is in an increasingly weird place. Sure, you could add an out-of-focus bee to an image in Photoshop for a long time. But it’s been a while since I’ve had the ability to do this. The same device used to take and distribute photos That’s another story: the power and accessibility of generative AI tools is outpacing our shared understanding of what’s real and what’s fake when scrolling through Instagram.
Personally, I find this feature the strangest when showing it to young children. They will grow up knowing that they can turn a rough sketch into something more polished with the push of a button. Or, with just a little effort, they can spice up a photo of railroad tracks by adding a train. Is this a good thing or a bad thing? I don’t know, but there’s definitely a dissonance between my perception of a work of art as a child and the way my child sees it.
None of this stops you from having fun making images from your sketches: There’s a seriousness to the generative AI’s output that makes you laugh a little, like when I tried to add a green monster poking its head out of Puget Sound, and the AI interpreted my drawing as a giant green polar bear with rippling muscles standing on the shore, or when it turned my stick figure sketch into a life-sized stick figure with a shadow on the ground.
Is the definition of photography changing before our eyes? Is our understanding of the truth in an image shifting during this incredibly precarious time for our democracy? Yes, but then you take a picture of a rabbit and an AI puts a little top hat on its little head. What times to be alive.
The sketch-to-image feature is available on the Galaxy Z Fold 6 and Z Flip 6. Samsung hasn’t said whether it will make the feature available on other Galaxy smartphones, but given the company’s track record of aggressively expanding Galaxy AI to previous generation models, it seems very likely. Samsung has also committed to bringing AI capabilities to 200 million smartphones this year alone. If that slightly blurry bee is any indication, things are bound to get a little weird when it does happen.
