For instance, if I post a comic, I usually describe what characters are involved and what they say in each panel, possibly adding expressions and gestures. But I won't describe their position in each frame, what they're wearing, their body shape, their hair color, etc. (unless it's a relevant fact in the comic itself).
A screenshot's description may end up different depending on whether I'm showing off my desktop or whether it's to pull a joke.
I recently made a joke response with Mastodon's user menu and my cursor hovering the "Block (user)" option. And would you look at that, the description I just gave you was pretty much the alt text! I didn't describe what the user's profile image was, what their bio says, or what other menu items there were, because it didn't matter.
But if I'm showing off my desktop, I won't laser focus on one part of it. Instead, I will give a rough overview of everything shown, but I won't describe the UI layout of all programs in detail, and I especially won't repeat all the labels. I'll only provide more details for parts I want to direct people's focus on.
I would write something like
"Screenshot of my desktop, showing the taskbar at the bottom and a music player on the right half. The wallpaper is a digital painting of a meadow in Autumn.
The taskbar shows the current time and some music playback controls on the left, an icon list of all running applications in the center, and small status icons on the right.
The music player shows a playlist and is currently playing (song) by (artist)."
I hope this post helps you in some way.