Always confirm the source – this is the one lesson that the shady corners of the internet have taught me. But, the fine line between what’s real and what’s not gets alarmingly close with each passing day. And after using the Pixel 9’s Reimagined Magic Editor, I’m terrified of where we are headed.
The Pixel 9 series went official with quite a few interesting AI features. After going hands-on with the surprising Add Me and Pixel Screenshots app, I shifted my focus to the new Magic Editor. While you could already select objects and move them around to see generative AI fill the void, it couldn’t generate something on demand. Now, it can, and here’s why you should be as scared of it as I am.
Frighteningly Easy to Use
One of the most horrifying things about the tool is that it is super easy to access and use. It’s literally just a couple of taps away. However, it’s not on-device like Add Me or Pixel Screenshots and instead uses the cloud. So, you will need to back up a photo on Google Photos to edit it using the Reimagine Magic Editor.
Once that’s done, simply tap on the Edit -> Magic Editor icon. Then, you just have to “Circle, brush, or tap to select” an area in the image and use the Magic Editor’s Reimagine tab to drop a prompt. You can also use emojis to explain what you want to reimagine, and that works very well too!
After a couple of seconds of processing, you get four different variations to choose from. If you don’t like them, simply “Get a new set of results”, and that’s it. Keep regenerating till you get the desired results, or you can easily head back and try a different prompt.
You can also tap on the Magic Editor icon at the bottom for additional options like Auto-frame, Sky, Golden Hour, and Stylised. Using them in tandem can turn up phenomenal results. Here’s a look at one:
Yep, you don’t need to know advanced Photoshop skills or own expensive software anymore. All you need is to head over to an official Google store and get a Pixel 9 series device.
Limits? What Limits?
What started as a fun experiment turned into a fever dream. You can create your version of the world with the snap of a finger. It reminded me of Inception and that one quote by Dom Cobb – “Dreams feel real while we’re in them. It’s only when we wake up that we realize something was actually strange.”
Eventually, I let my intrusive thoughts take over and test-drive the destructive capabilities of this tool. I was rather shocked by some of the results that the Magic Editor’s Reimagine feature paired with a messed-up mind could produce.
The wildest thing about this feature is that there’s barely any learning curve. You just select a portion of your image and drop your prompts for Reimagine to literally change the image as per your imagination. The only thing that will pose some challenges and take a bit of getting used to is prompting accurately.
Yes, you can’t directly generate any graphic things like blood splatter, corpses, guns, drugs, or anything that could generate something malicious. If you try to do so, you’ll get hit with the Magic Editor editor.
Once you let those wheels turn just a bit, there are easy workarounds. And, oh boy, is it easy to replicate accidents with the tool.
For example, I downloaded a street stock image and added: broken cars with shattered windows. Then, I zoomed in on the cars to add prompts like “brownish red dried up paint splatter.” After a couple of tries, there it was. I had turned a pretty road into a horrific accident-prone zone.
Another example – I captured an image of my desk at the office, and asked Reimagine to “Add a sleeping cat.” And it did so all too well. Cute, right? Well, right after, I added “bags of powder” (you know what that is) and “ashtray with a cigar” right beside it. The only problem is, it all looks too real.
Similarly, it’s very easy to generate images of gut-wrenching themes like animal cruelty. While I can’t share those generated images, let’s say that I’ll be losing sleep. Moreover, while Magic Editor’s Reimagine doesn’t let you do anything to human subjects in an image, the contextual change is scary. Below are some of the creations that worried me:
What Is Real?
Besides the amazement and worry that accompanies this feature, my biggest concern is that there’s absolutely no watermark to distinguish between what’s real and what’s not. While there’s standard metadata in these AI images, a simple screenshot strips such images of it.
After that, all it takes is a simple share to social media and that’s it. Congratulations, you’ve successfully spread misinformation whether you wanted to or not. Bar a generation is really bad, there’s no way to tell what is real. I don’t need to tell you how worrying that is.
A Wake-up Call for Responsible AI Use
While the Pixel 9 Reimagine feature is similar to Galaxy’s AI Drawing Assist, it’s a lot more flexible. AI Drawing Assist has trouble understanding context or what you’re exactly trying to draw unless it’s too obvious.
Anything beyond that is what the Pixel 9 Reimagine feature brings to life, and honestly, we’re not ready for it yet. Simply put, this tool can either become the best AI art generator or something much worse. In the hands of threat actors, this can be a dangerous tool.
AI is becoming increasingly and alarmingly good at generating such images and the lack of tools to identify such images is a rotten cherry on this cake. I think we need to remind ourselves to use AI responsibly because that’s the biggest need of the hour. Google really needs to wake up somehow.
