logoalt Hacker News

estearumtoday at 4:36 PM1 replyview on HN

Quite telling that you think a technology that merely prevents you from passing off an AI-generated image as not-AI-generated makes the model "worthless."

Good!

That's the point! Whatever amazing use case you had in mind is bad and I'm glad SynthID (apparently) makes it impossible.


Replies

nightskitoday at 5:18 PM

Actually no it just makes me use a different model. My uses are not nefarious at all, although it's fine for you to assume so. There are real, legitimate reasons why SynthId is actively harmful that do not involve deceiving or manipulating people at all. SynthId is just a stain on legitimate AI users. People who want to deceive or manipulate are not using Google models anyways. They are going to use a model without safety rails (which is not what I am advocating for per se, just that SynthId is an awful solution).

It actually reeks of Google, since it's a technical solution to a people problem. Google doesn't seem to understand people.

show 2 replies