I made a post a while back when OpenAI released their GPT image update about how this would change thumbnails...
Update: it has cut down the amount of time I spend on thumbnails by a LOT and can do things I don't have the photoshop skillz for.
I'm not generating entire thumbnails in one shot- I'm creating assets that would normally take me a lot of time scrolling through Google Images for, or bringing ideas to life that I can't make from scratch.
Attached below: this was a 1 of 10 for us and I generated the broken apple logo in GPT after visualizing it in my mind but not having the skills needed to make it from scratch in Photoshop.
It took maybe 2 prompts and 45 seconds for GPT to create exactly what was in my mind's eye.
Another example: I wanted a stock image of a hand holding an iPhone in a really particular angle that I was having trouble finding on Images, so I had GPT generate it for me. Furthermore, I had it also do a lot of the lighting effects that I normally spend too much time hyper-fixating on. Also, the mask in the background at the exact angle I wanted.
Generating both assets took about 5 minutes AKA way less time then it would've taken me to actually find the images on Google and also do all the lighting/extra FX.