I recently watched a podcast that discussed the growing use of AI by some YouTube creators. One of the people involved in the discussion mentioned that they had evidence that some of their content had been copied without their permission and then reworked using AI before being uploaded. The plagiariser then earned income from the ripped off work thanks to the YouTube algorithm.
Another topic that was mentioned was the use of AI to create the thumbnails that creators upload to YouTube to attract viewers ... and that they all tended to look the same.
I decided to see what sort of job ChatGPT would do if I asked it to design a thumbnail for my most recent YouTube video. The thumbnail I created using the opening image of my video looked like this ...
... and ChatGPT's design looked like this:
I must admit that the end result is quite impressive ... but I'm not sure if I'll be asking ChatGPT to design my YouTube thumbnails in the future.
I liked your A.I. generated thumbnail, but it also shows that it would just blend into the YouTube crowd...
ReplyDeleteRoger,
DeleteThe problem with AU- generated thumbnails is that- as you comment - that they can all end up looking the same.
All the best,
Bob
Bob, the sameness point is useful: a polished image can still lose the details that make a particular video recognisable. Keeping a real frame from the museum visit as the starting point seems especially valuable here. Disclosure: I work on Minecraft Thumbnail Maker, a paid AI thumbnail tool for a different gaming niche. Even there, I think the actual build or scene should remain the reference, and the creator should check the result against the footage rather than treating polish as proof of accuracy.
ReplyDeleteRichard did,
ReplyDeleteThanks very much for your comment.
I do use ChatGPT to help polish images that I’ve created, but not to create them from scratch. The latter seems to feel ‘wrong’ to me … and I’ll be continuing to use my own photographs as YouTube video thumbnails.
All the best,
Bob