I don't think the AI knows what words or text are. It just knows some things have these angular line arrangements in certain spaces. It doesn't care if the text says something, just that it superficially resembles what it thinks text looks like.
Very different, you could train one to be really good at making signs, but you'd need a huge dataset of images of things with text and transcriptions of the text in the image.
Otherwise you just get this emulation of the general idea of text that's contained in images that look like the thing that you're asking the AI to replicate by smooshing all the similarly tagged images together.
I don't think the AI knows what words or text are. It just knows some things have these angular line arrangements in certain spaces. It doesn't care if the text says something, just that it superficially resembles what it thinks text looks like.
Would it be hard to integrate somekind of text transcriber into one of these
We already have stuff that can transcribe stuff from pictures and ai text engines already exist
I don't know enough about this stuff to know how closely related those two things are, an AI that does images and one that does text.
Very different, you could train one to be really good at making signs, but you'd need a huge dataset of images of things with text and transcriptions of the text in the image.
Otherwise you just get this emulation of the general idea of text that's contained in images that look like the thing that you're asking the AI to replicate by smooshing all the similarly tagged images together.