Eventually, you won’t be able to and we’ll be back to what my generation learned early on, don’t trust what you see on the Internet.
lmao. the circle of life.
Did you worry about photoshops too? Because many of those were absolutely hard to tell
Not true. If it looks shopped I can tell from some of the pixels and from seeing quite a few shops in my time.
Be careful of falling victim to bad toupees.
You always know when you spot a bad fake, but you have no idea how many good fakes flew right past your radar. We tend to think we’re much better at these things than we really are.
You’ve never seen an expert
Go see art in real life. It’s all original content.
Watermarking never meant to be a legitimate way of telling this stuff apart. Anyone slightly savvy could run a generator locally and produce images without watermarking, if they really meant to fool others. You are right in that the watermarking makes it (or did) more difficult for the average joe, but yeah. This is a never was kind of scenario
We can’t tell the difference anymore - we simply can’t trust anything on social media anymore as it’s mostly not true - also print and broadcast media are equally lies…
Always have been.
I drop musical artists like it’s hot the moment I find out they support Israel. I’m not opposed to doing the same for AI.
We don’t… That’s the new reality we live in and it happened so fast my head hurts
If you can establish that it was written before 2021 or so, it’s probably not from an LLM. That kind of text has been compared to “low background steel”, typically salvaged from ships sunk in WW2 or earlier. Low background steel is a valuable commodity used for some types of highly sensitive instruments. Steel made in August 1945 or later, especially post 1950-ish, always has trace amounts of radioactivity from u-know-what.
Note that contamination is contagious. Even if you know that a human wrote something post-2021, for example because you wrote it yourself, you still can’t say it’s guaranteed AI-free, because of how AI has warped language and to some extent everyone’s minds.
https://englishlanguagestudies.com/the-impact-of-artificial-intelligence/
https://www.psychologytoday.com/us/blog/urban-survival/202507/the-emerging-problem-of-ai-psychosis
A.I. has quietly added quietly to everything. You can tell if a YouTube video script has A.I. assistance if they say quietly a lot. I dunno why it’s obsessed with that word. Sneaky fuckers
I didn’t know about “quietly”. It used to be “delve”. Bah.
It’s fundamentally impossible to watermark digital data permanently without controlling the hardware so the only answer here is to use brain.
I suspect hardware watermarking market will emerge as those types of watermarks are much harder to remove. I.e. the phone camera uses cryptography to sign that this picture was taken by a real phone. So proof will be inverted and that might actually be a good thing.
But brain no good! Good brain bad!
Reverse image search? If it had a watermark then it’s probably been posted somewhere before. If it hasn’t, then no you can’t really know. You can get a sense for common AI styles and features, but unless the original artist posted it somewhere, in certain cases you’re just guessing.
If it’s good enough to not know it’s ai then it’s good enough to not be called slop
