Skip to content
The State of AI
Images & photos
Wondering what’s coming next for AI? and what they mean for product managers—both for our roles and for the products we build.
Generating pictures.
When DALL·E 2—Open AI’s image generator—was released in 2022, I couldn’t stop testing it to see what it would come up with for different prompts. The images it was creating then were impressive, but they’ve become even more so, especially ones including text.
For example, in 2022 a asked recently-released to create a very meta “DALLE-2: Revenge of the AI, pixar studios movie poster.” While at first glance, the poster looks passable, when you look closer you’ll notice the strange text, the “noise” or irregularities in the background, and that it doesn’t really look like Pixar’s style. I tried the same prompt with (released in August 2023), and the result is much better: believable imagery in the right style, and a tagline that actually makes sense.
Prompt: “DALLE-2: Revenge of the AI, pixar studios movie poster.”
Generating “photos”.
When image AI models first arrived on the scene, they were notoriously bad for generating slightly uncanny-looking people and extra fingers, as shown from ZDNET in 2022. But now, using that same prompt—“photo of a person with glasses making a point to several people at a conference table in a meeting room”—with DALL·E 3, we get something much more accurate and almost impossible to distinguish from a real photo. No extra fingers here!
Prompt: “Photo of a person with glasses making a point to several people at a conference table in a meeting room.”
Want to print your doc?
This is not the way.
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
CtrlP
) instead.