Samsung’s Fresh AI Image Generation Tool Is a Little Too Good

Share

The pirate ship in Elliott Bay was one thing, but it was the slightly blurry image of the bee that unsettled me.

Samsung would very much like us (and its shareholders) to know that its novel phones are the most AI-powered phones ever, and the Fold 6 I tested comes with a novel tool called “sketch to image.” Draw a sketch on a photo or blank note page, and the device uses generative AI to turn that thing into an image. I dismissed it as just another AI thing when Samsung announced it on stage at Unpacked — but, man, is it really good. So good, in fact, that it worries me a little.

Using the sketch tool on a note’s image is pretty harmless: you draw something, select it, and choose from a few styles, like “3D cartoon” and “illustration,” to turn your drawing into something more detailed. Your image is uploaded to the cloud, and after a moment, you’ll see a few options to choose from. The results are usually cute and entertaining; I’ve taken requests from my two-year-old and drawn goofy-looking dump trucks and school buses. You sometimes get a teddy bear with too many arms, but nothing grave.

Using a sketch to paint over a photo is where it gets weird. I’m the worst artist in the world, and this tool turned my very basic sketches into photorealistic images. The AI-generated elements are convincingly incorporated into the photos — scaled and aligned with the surroundings in a way that makes them tough to spot as fakes.

That’s how I came across the bee problem. I took a photo from a dock south of downtown Seattle with a few flowers in the foreground. Because they were close to the camera and I was focusing on the distance, they’re slightly blurry. I drew the world’s worst sketch of a bee on one of those flowers, thinking the AI ​​would insert a pointed image of the bee—easily giving it away as a counterfeit. Wrong!

The AI ​​bee is as blurry as the flower it lands on. If I didn’t know the origin story of the AI ​​bee, I wouldn’t think twice about scrolling through this image on Instagram. I would assume the photographer had taken the shot at the perfect moment, or was waiting for the bee to fly into frame—things that require skill and patience. Not true. In fact, I’m not even sure I would have noticed the “AI-generated content” watermark in the corner of the image.

At first glance, it looks convincing – but if you look at it for more than a second, you’ll notice something’s wrong

I’ve been playing around with sketches a lot in the past week to create an image, and the results aren’t always “bee-fuzzy” good. They often have the telltale signs of generative AI art—words scribbled in a foreign language, or weird textures that don’t quite look right. Convincing at first glance, but if you look at them for more than a second, you’ll notice something’s wrong.

Sometimes the content itself gives it away—I don’t think anyone would believe I saw a giant pirate ship anchored in Elliott Bay or a giant orange cat at a West Seattle intersection. But even when the images are so bizarre that no one could mistake them for the real thing, they look real.

Generally, the gigantic stuff will look obviously counterfeit. But it’s very simple to add another car to a photo of a busy road or a sailboat in the distance, and most people won’t know it. Other than that AI watermark — which can be easily cropped out — there’s no way to know there’s anything unusual in the photo. It’s weird!

The fuzzy bee will not destroy the fabric of our society

I don’t want to overstate this. Using Sketch to an image is completely optional, and many people will never find it in a gallery app. A blurry bee isn’t going to destroy the fabric of our society. But I think we’re in an increasingly weird place with AI. Sure, you’ve been able to add blurry bees to an image in Photoshop for ages. But putting that ability in the exact same device you exploit to take and distribute photos is another matter. The capabilities and availability of generative AI tools are outpacing our collective understanding of what might be real and what might be counterfeit when you scroll through Instagram.

Personally, I feel weird when I show this feature to my little one. They grow up knowing that with the click of a button, you can turn a sketch into something more elaborate. Or, with a little effort, you can spice up a photo of train tracks by adding a train. Is that a good thing or a bad thing? I have no idea, but I definitely feel a dissonance between how I viewed art as a child and how he will view it.

None of this stopped me from having a lot of fun sketching a picture. There’s a sincerity to the results of generative AI that’s quite entertaining—like when I tried to add a green monster poking its head out of Puget Sound and it interpreted my drawing as a giant green polar bear with rippling muscles standing on the shore. Or when it turned a stick figure sketch into a life-size stick figure with a shadow on the ground below it.

Is the definition of photography changing before our eyes? Is our understanding of truth in images changing in an incredibly uncertain time for our democracy? Yes, but I took a picture of a rabbit and AI let me put a diminutive top hat on its head. What times to live in.

Image Sketch is available on the Galaxy Z Fold 6 and Z Flip 6. Samsung hasn’t publicly stated whether it will bring the feature to other Galaxy phones, but given Galaxy’s history of aggressively expanding AI to previous-generation models, I think it’s extremely likely. Samsung has also committed to bringing the AI ​​feature to This year alone, 200 million phones were sold. If the little fuzzy bee is any indicator, I’d say things are bound to get a little weird when this happens.

Latest Posts

More News