I’m annoyed, not for the first time, by something Sam Altman has said. But this time it’s because I’m annoyed at how much I agree with what he’s saying — even though I think his statement is kind of bullshit.
In a recent interview, journalist Cleo Abram asked Altman how people will be able to tell what’s real and what’s not in an age of convincing AI-generated content. Specifically, she references the bunnies. You know the ones I mean: caught in some Ring-camera-ish footage of a backyard, discovering and jumping on a trampoline. So cute! So wholesome! So completely AI-generated! The video went viral, almost certainly before the people liking and sharing it realized that it was AI.
So what happens as the technology improves and AI content is everywhere? How will we know what’s real? Eh, we might not, Altman seems to say. By way of an example, he points a finger at something I spend a lot of time thinking about: phone cameras.
“Even a photo you take out of your iPhone today, it’s like, mostly real but it’s a little not,” he tells Abram. So much processing happens between the photons hitting an image sensor and the final image, he says, and what we end up with is a kind of optimized version of reality. And sure, he’s right. Any old digital camera makes a million decisions about a scene: contrast, sharpness, which pixels should be red and which ones are green. A phone camera goes a lot further, combining data from different frames, deciphering what’s the ground and what’s the sky, and brightening faces to look a little more flattering.
Altman’s point is that we accept this level of manipulation as “real,” even though we know there’s more going on. As AI content becomes more commonplace, “I think the threshold for how real does it have to be to be considered to be real will just keep moving,” he says. That’s when I started clenching my jaw.
For starters, there’s a big difference between a photo that starts with photons hitting a sensor and one that is fabricated from scratch with generative AI. If they exist on a spectrum, then it’s pretty god damn big spectrum. I also think that most people aren’t aware of what kind of processing happens when they take a picture with a phone, and it’s not as liberal as his statements suggest. Your iPhone camera isn’t in the business of changing details or adding things that weren’t there. Even when it seems to be doing something screwy, the explanation is usually pretty simple. Sure, sometimes demon face happens. AI moon is one thing, and you can get wild with Google Photos’ gen-AI editing tools. But I haven’t seen evidence of the cameras themselves going rogue and adding elements that weren’t there in the last five years of testing every major phone on the US market.
Invoking phone camera processing as an example of an acceptable un-reality is annoying, but I think Altman is generally right. Our understanding of what is real and what isn’t changed when Photoshop took off. I know all kinds of staging and editing goes into a magazine cover photo, but I still accept a picture of Sarah Jessica Parker on the cover of Vogue as “real.” This understanding has already changed in the AI era when we look at a picture on social media, or an ad, or a product listing — and that will only continue. But Altman implies that as our definition of “real” or “real enough” changes, we’ll appreciate it all the same as something we see with our own eyes. After all, we enjoy science fiction movies even though we know they’re not real, he points out.
But I think that we’re still going to care whether something is real or not and calibrate our enjoyment accordingly. The video of the bunnies on the trampoline is so much less enjoyable when you know it’s not real. The whole premise of the the thing is “look at the funny thing these rabbits did.” That’s only funny if it’s real! If social media becomes awash with cute but unreal videos, I don’t think I’m going to stop caring and just enjoy them. I think I’m going to stop enjoying that social media app. Who knows? Maybe I’ll be spending more time with an LTE smartwatch and less time with my phone in the future.
