site banner
PaperclipPerfector  14hr ago (text post)   497 thread views

Friday Fun Thread for October 18, 2024

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

If you haven't seen it yet: Scott's AI Art Turing Test. See if you can guess which pictures are AI and which are human.

EDIT: Forgot to mention, the quiz won’t automatically tell you your score at the end, so if you want to know your score, you’ll have to write down your answers and manually compare them with Scott’s answer key.

Spoilers below where I discuss some of the answers, don't look until you've done the test yourself:

I did pretty terrible! I said they were almost all AI, which was definitely not correct. I got the impression after the first few that Scott might have been pulling a trick and he made them all AI to see how people would react. I did continue to analyze each individual image though, and I did feel that there was a legitimate case for almost all of them being AI. The three that I said were human - Tropical Garden, Creepy Skull, and Flailing Limbs - did indeed turn out to be human, so that's good. If I'm going to err, that's how I would prefer to be calibrated. I was pleased to find out that the best painting in the set, Saint In Mountains, was human. But oddly I couldn't find that exact version of the painting anywhere else on the internet using Google image search. Scott's version looks like it had a color/contrast filter applied to it compared to the version on Wikipedia (search for "Saint Anthony Abbot Tempted by a Heap of Gold") and every other version I could find. Scott acknowledged that he cropped some of the photos, but he didn't say anything about adjusting contrast or making any other edits. If he did make any edits like that, or if he simply picked a more uncommon version of the painting, then that could definitely bias the results. Part of what tipped me towards saying that Saint In Mountains was AI was the way that the thick black section at the bottom of the cloak looked unnatural. If I had seen the Wikipedia version, where the detailing on the fabric is more visible, I would have been more likely to identify it as human.

I was pretty deep into image generation at one point and got the following wrong

  • Fancy Car, pretty obvious in hindsight but when I was looking at it I convinced myself that the headlights and wing mirror had some weird distortions.
  • Rainbow Girl, the way the artist chose to render the hair on the right side of the image is frankly bizarre and I thought the ear was a bit odd.
  • Giant Ship, decided it was AI at a glance based on the subject matter and rendering style. Very obviously CG/photobash if you know that frontier models still struggle with ship rigging and other similar linear patterns.
  • Still Life, didn't notice any obvious tells
  • Paris Scene, thought I recognised this. Even knowing it's AI there aren't many things I can point to as obvious tells.
  • Pretty Lake
  • Colourful Town, wrote this off as AI because I couldn't make sense of the composition

Many correct guesses were with very low confidence. Stylised landscapes and certain outdoor scenes may as well be a coinflip.

I think I got about 80% right. I'm a bit miffed that Riverside Cafe, which I quite liked (even though it looked like hotel-grade art), turned out to be AI.

I basically marked all hyper-detailed or overly colorful paintings as AI (with the exception of the first ship, because it was right next to the other ship, so I scrolled back and saw that it all made sense).

The oil paintings I tried to classify based on their composition and perspective.

The abstract paintings were the hardest, since I had to resort to thinking about the emotions of the artist. Surprisingly, I got all of them right.

The anime girls were the easiest, because the AI-generated one had that face. It's like there's one AI model in existence for drawing anime girls, and everyone uses it. It's like Lobsters typeface, once you've seen it, you start to notice it everywhere.

P.S. @ZorbaTHut, aren't spoilers supposed to be inline?

The key to the mother and child picture is that the child is missing the halo, and any depiction of a regular saint would be more differentiating from Mary

I'd say that I did pretty well, maybe around 80% accurate from eyeballing the results.

The ones that really tripped me up were the abstract paintings (and the ones 'intentionally' made by humans can often look like an explosion in the paint factory in the first place) and some works in an older style. There was a human painting in the last 8 that had really fucked up hands and feet on the dozens of characters, so that threw me for a loop.

Of course with each passing day, it gets harder and harder to tell, and Flux is absolutely solid, though I haven't had the luck of using it yet.

The ones that really tripped me up were the abstract paintings

Yes, those are a coinflip at this point. Occasionally there are still tells though, like with Bright Jumble Woman, if you zoom in on the eyes you'll see artifacting that is very characteristic of AI so I was confident on that one. But with something like Purple Squares, there's no way of knowing.

I'm pretty sure that the test specifically asked us to avoid zooming in haha.

But I was also looking out for artifacts. A lot of smaller details lack symmetry and become squiggly in a nonhuman way. Also random blotches and spots that don't make much sense.

Can we see the answers without being a subscriber?

The answer key is the first comment on the post

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats