Reply to thread

Message

<blockquote data-quote="J.Quondam" data-source="post: 9156007" data-attributes="member: 7030100">As I understand it, the signature issue is an outdated problem at this point. It's a known serious issue, and is being actively corrected. It will be engineered away soon, if it isn't already.Anecdotally, I've been playing with a free online image generator built on a recent rev of the latest generation AIs, DALL-E 3 and ChatGPT4. I've churned out hundreds of images and I've seen thousands of others, yet I haven't spotted a single art signature or stockphoto watermark. (I've never asked for one specifically, though, which would be an interesting test.) A year ago and with any (free) AI tool, those marks appeared on most any image generated, like a pox. But they're almost eradicated now, afaict.Many technical issues that threaten to stir up big legal problems (eg, training consent, copying signatures, regenerating IP, etc), are mostly identified pretty early on, and then fixed ASAP thereafter. They have to, because it would be legally stupid not to fix these issues quickly.*So... what issues are NOT going away in the next version of the &quot;AI&quot; app? The ethical ones regarding compensation and jobs; creator power vs corps; credit and recognition; safety vs choice; free speech, fair use vs IP; privacy; AI regulation; and so forth.* Of course, it would be better not to deploy a tech willy-nilly without fixing issues before they become a problem. The techbro mantra &quot;It's easier to ask forgiveness than permission&quot; needs to end.</blockquote>

[QUOTE="J.Quondam, post: 9156007, member: 7030100"] As I understand it, the signature issue is an outdated problem at this point. It's a known serious issue, and is being actively corrected. It will be engineered away soon, if it isn't already. Anecdotally, I've been playing with a free online image generator built on a recent rev of the latest generation AIs, DALL-E 3 and ChatGPT4. I've churned out hundreds of images and I've seen thousands of others, yet I haven't spotted a [I]single[/I] art signature or stockphoto watermark. (I've never asked for one specifically, though, which would be an interesting test.) A year ago and with any (free) AI tool, those marks appeared on most any image generated, like a pox. But they're almost eradicated now, afaict. Many technical issues that threaten to stir up big legal problems (eg, training consent, copying signatures, regenerating IP, etc), are mostly identified pretty early on, and then fixed ASAP thereafter. They have to, because it would be legally stupid not to fix these issues quickly.* So... what issues are NOT going away in the next version of the "AI" app? The [I]ethical[/I] ones regarding compensation and jobs; creator power vs corps; credit and recognition; safety vs choice; free speech, fair use vs IP; privacy; AI regulation; and so forth. [I][SIZE=3]* Of course, it would be better not to deploy a tech willy-nilly without fixing issues before they become a problem. The techbro mantra "[/SIZE][/I][SIZE=3]It's easier to ask forgiveness than permission[/SIZE][I][SIZE=3]" needs to end.[/SIZE][/I] [/QUOTE]

Verification