Issue #59: What is a Photo? — Jason Michael Perry

Howdy👋🏾. The power of AI has blurred the lines between reality and make-believe in ways that many are unprepared for. One glaring example is the rise of deepfakes—advanced technology that can create lifelike but entirely fabricated audio and video content. We’ve already seen this tech rear its head in the current Presidential election, with fake robocalls mimicking President Biden using voice cloning tools.

Closer to home, in Baltimore, a high school principal experienced the devastating repercussions of this technology firsthand. An offensive audio clip leaked to the media was a deepfake. Unfortunately, the truth came out only after he had served a lengthy suspension and was subjected to national scrutiny for something he didn’t do.

While deepfakes are a significant concern, a quieter but equally important question is emerging: What is a photo? As a kid, I remember watching detective shows where the case was cracked wide open by photographic proof—a stack of recently developed photos in a manilla folder, indisputably proving the show’s guest star’s innocence. Back then, a photo was definitive proof of reality.

But cameras have changed; today, most “photos” are taken by smartphones. Our social media feeds and text messages are filled with images snapped on these devices, which we quickly label as photos. The truth is, as technology has advanced, these images are rarely true representations. They’re not just photos anymore; they’re digital composites. Your phone’s software often combines multiple shots in rapidly, ensuring everyone’s eyes are open, smiles are perfect, and the lighting is just right.

Like many, I’ve appreciated these adjustments, even if the final product is more a digital composite than a traditional photo. After all, it still captures a moment that happened, right? Right.

The original image is on the left. The center image is edited with RunwayML. Right image edited using Google Photos Magic Editor

These tools in our “cameras” are getting increasingly powerful, and adding AI-powered features makes it even tougher to identify truth from fiction. Just two weeks ago, Google released its new Pixel phones, packed with incredible AI features, including “Add Me.” This feature allows a person to be inserted into a group photo, even if they weren’t there when the photo was taken. Of course, tools like Photoshop have offered this capability for years, but it requires skills and knowledge that even the royal family struggled with. Now, Google has made it simple for anyone to create these digital compositions. Bob may have missed the first few days of the conference, but he’s still in the group photo on social media taken on day one.

And that’s just the beginning. Many of these camera tools are slowly adding features that allow AI enhancements to photos—changing backgrounds, adding objects, or even fabricating entire scenes. The Verge has an excellent series where they modify real “photos” to include fantastical elements, challenging our perception of what’s real.

This evolution forces us to ask: When you look at a photo or a digital composite, should you consider it a reality? More on this in a bit, but first, my thoughts on tech & things and a word from our sponsors:

🧠 Over Half of Fortune 500s List AI as a Threat
More than 56% of Fortune 500 companies now list AI as a risk factor in their SEC filings, highlighting the transformative impact of AI on business landscapes. Companies slow to adapt could face significant challenges.

🖥️ Don’t Connect Your TV to the Internet
TV manufacturers are increasingly turning your TV into a data-collecting and ad-serving device. This article dives into how connecting your TV to the internet is more about making money than enhancing your viewing experience.

🎮 Epic Game Store Launches in the EU
The EU’s DMA is finally allowing competing app stores like the Epic Game Store to operate. With Fortnite making a comeback in the EU, explore what this means for app marketplaces and the future of gaming.


The original image is on the left. The center and right images are generated using Google Photos Magic Editor.

So, where do we go from here? One place to start is by being more precise with our language, ensuring we label things accurately. Nowadays, it’s increasingly rare for a “photo” to be an accurate, untouched representation of reality. When you use most cameras, including those on your phone, multiple images are generated:

  • RAW: This is the unprocessed image directly from the camera’s sensor, with no adjustments or corrections.
  • Processed Image: We typically interact with this—a version of the RAW image enhanced through various adjustments like color correction, sharpening, and more. We often combine multiple RAW files to create the best possible image.

From here, further modifications can turn an image into a digital composite—a term for an image that has been extensively edited and manipulated. A lighter form of this is the retouched image, which focuses on making the subjects look more polished.

Based on these definitions, only the RAW file—an unaltered, unprocessed image—truly meets the definition of a photograph, or “photo” in the traditional sense.

Unfortunately, most non-professional photographers don’t keep the RAW files. The processed images are often preferred because they look better and have enhancements that make them more appealing than the raw originals.

For many of us, especially businesses, the burden of proof when it comes to images must become much higher. You should assume that anything you see has had some level of adjustment or enhancement. It’s crucial to start viewing these images not as definitive proof of reality but as representations—possibilities rather than certainties.

This leads to a critical question: Should an insurance company accept a camera photo you take as evidence? Should courtrooms continue to rely on photographic evidence as indisputable?

One possible solution is to increase the use of watermarking and other technologies that clearly label what these images really are. But for now, we must start asking: Is this an authentic photo?

-jason

p.s. I’ve often wondered where self-driving cars go at night, and it looks like a neighborhood in San Francisco has also learned. Several go to a parking lot in the evening but apparently get stuck honking at each other, keeping an entire neighborhood awake in what feels like a digital version of Who’s on First or the Three Stooges. Waymo says they pushed updates to fix these issues, but neighbors say the honks are not gone yet.