Beth

So the image...

is everything.

And everyone's really excited that the text in an image is accurate, meaning that it's actual text.

It's not spelling things wrong in the same way that it used to be.

It hasn't made up words in the same way that it used to be.

So it's rendering text well, but then you can't edit it.

Now, if you go to Canva and say grab text, right, there's a grab text edit on Canva, then it's going to convert it to actual text that you can correct.

But the other piece of this for me is taking a step back

in terms of how accessible AI makes information in like a ton of languages.

And that's also not gonna happen.

I mean, it may be that if you say, okay, Google slide,

I've corrected all of these in my image generator, in my image editor.

I've gone in and changed the text and all that.

It will generate you a new image with text in another language.

But I do think that we've taken a step back about how accessible content and information is.

Well, it's one of the... You don't understand where I'm coming from, Beth.

Gareth says...

Well, I would say that my point is that putting everything in an image and needing to use an image generator instead of text, like text being able to edit text or the text overlay,

means that it's a different process, right?

Generating images takes more power, takes more skill, creates a lot of opportunity for something to get built as a mistake.