AI Quiz

[Joyager2]

I know most AI threads devolve into miserable little arguments, but the NYT put out this short quiz yesterday that puts AI writing side-by-side with excerpts from classic writing. I’m curious to hear which passages you preferred to read and why, or how you knew which was which (if you knew at all).

Personally, I chose the human writing all down the line. I think that’s in part because I’ve read a lot of the books these excerpts are from and could recognize them immediately as stories that I’ve read and loved and knew the context for, but also because there’s a pretty clear pattern in some of the prose that Claude put out (there may also be some concern here regarding the fact that the NYT only used one AI) that makes it sound shallow and careless. But the data says a lot of these choices were split, so I wonder what other folks felt about the human writing that made them feel that an AI might have sounded better in certain situations.

Edit: Most folks seem to think I'm looking for whether or not you can tell which excerpt is AI and which one's not. That's not really what I'm going for here (and not what the article is about, either, I think). I'm mostly trying to figure out whether or not, before you know which one is which, you have any inclination away from the human writing or towards the AI writing. And also why. I'm hoping to talk about the way the prose works (or doesn't work) here, not necessarily how well you are at recognizing AI writing.

 

[Author_Riceball]

I know most AI threads devolve into miserable little arguments, but the NYT put out this short quiz yesterday that puts AI writing side-by-side with excerpts from classic writing. I’m curious to hear which passages you preferred to read and why, or how you knew which was which (if you knew at all).

Personally, I chose the human writing all down the line. I think that’s in part because I’ve read a lot of the books these excerpts are from and could recognize them immediately as stories that I’ve read and loved and knew the context for, but also because there’s a pretty clear pattern in some of the prose that Claude put out (there may also be some concern here regarding the fact that the NYT only used one AI) that makes it sound shallow and careless. But the data says a lot of these choices were split, so I wonder what other folks felt about the human writing that made them feel that an AI might have sounded better in certain situations.

Is this ai

 

[Corty]

@grokkingSmut summarize this for me as a smut novel
Also:

 

[BearlyAlive]

Good to know I still got it. Sniffed out those clankers at a 100% rate, but the science writing part almost got me, not gonna lie. Everything else was fairly obvious, in my opinion.

 

[Joyager2]

Good to know I still got it. Sniffed out those clankers at a 100% rate, but the science writing part almost got me, not gonna lie. Everything else was fairly obvious, in my opinion.

The science one almost got me too, but largely because I haven’t read any Carl Sagan in a hot minute. But I think, ultimately, the actual excerpt is just so obviously more complex (and therefore more interesting) that it becomes clear after a reread or two. It just feels like it’s actually saying something, whereas Claude does not. But 70% of readers apparently disagree.

 

[Nevafrost]

Btw, even if I 'accidentally' chose the AI writing, I would have hated the moment I knew it was AI. That's because it's "AI". Fck this shit.

 

[Joyager2]

View attachment 47344
Btw, even if I 'accidentally' chose the AI writing, I would have hated the moment I knew it was AI. That's because it's "AI". Fck this shit.

Can I ask which (if any) of the AI passages you chose? Ethical issues aside (big ask), I’m mostly interested in what parts of the prose itself either turns folks off of the human excerpts or makes them (in the moment, at least) prefer the LLM.

 

[laccoff_mawning]

The only one I chose AI for was the first one.

Which I'm completely okay with. Since an old crumbling building is much cooler than someone's weird take on war.

 

[Failnot]

100% clearance light work no reaction

 

[Conqueror_Quack]

I know most AI threads devolve into miserable little arguments, but the NYT put out this short quiz yesterday that puts AI writing side-by-side with excerpts from classic writing. I’m curious to hear which passages you preferred to read and why, or how you knew which was which (if you knew at all).

Personally, I chose the human writing all down the line. I think that’s in part because I’ve read a lot of the books these excerpts are from and could recognize them immediately as stories that I’ve read and loved and knew the context for, but also because there’s a pretty clear pattern in some of the prose that Claude put out (there may also be some concern here regarding the fact that the NYT only used one AI) that makes it sound shallow and careless. But the data says a lot of these choices were split, so I wonder what other folks felt about the human writing that made them feel that an AI might have sounded better in certain situations.

Fumbled with the science one, that's all.

 

[TheTrinary]

I know most AI threads devolve into miserable little arguments, but the NYT put out this short quiz yesterday that puts AI writing side-by-side with excerpts from classic writing. I’m curious to hear which passages you preferred to read and why, or how you knew which was which (if you knew at all).

Personally, I chose the human writing all down the line. I think that’s in part because I’ve read a lot of the books these excerpts are from and could recognize them immediately as stories that I’ve read and loved and knew the context for, but also because there’s a pretty clear pattern in some of the prose that Claude put out (there may also be some concern here regarding the fact that the NYT only used one AI) that makes it sound shallow and careless. But the data says a lot of these choices were split, so I wonder what other folks felt about the human writing that made them feel that an AI might have sounded better in certain situations.

This was interesting.

A little manipulative as context is hugely important and certain matchups felt intentionally obfuscating. I only missed Blood Meridian, but that is hardly my fault. It's meant to be clunky and sometimes awkward.

The general summary is that AI is saying not much very efficiently.

 

[Joyager2]

The only one I chose AI for was the first one.

Which I'm completely okay with. Since an old crumbling building is much cooler than someone's weird take on war.

So it's mostly the content for you in that one and less the actual prose? Or is the way Judge Holden speaks that got you to prefer the snippet about the church?

 

[rileykifer]

Just about every passage sounded like AI to me, but my brain is also a bit fried right now. My result ended up being no preference. (Though only three I chose the AI option for, and those were a "this bores me less" type of choice instead of actually liking it.)

 

[Joyager2]

Just about every passage sounded like AI to me, but my brain is also a bit fried right now. My result ended up being no preference. (Though only three I chose the AI option for, and those were a "this bores me less" type of choice instead of actually liking it.)

That's really interesting. Would you say it was the content of the excerpts or the writing that made you feel bored?

 

[laccoff_mawning]

I ended up judging by both content and prose. the "as well men ask what they think of stone" definitely deducted a few marks, but even if that sentence was replaced with something nicer on the eyes, I still think the content of the crumbling stone is better than war being described as the ultimate trade waiting for the ultimate practitioner. the entire thing just didn't ring with me.

Another example would be the science one. As soon as you call humans matter forged in dying stars billions of years ago I lose all interest in this being "scientific", thus I vote the other.

I guess I place a high emphasis on the philosophy of the writing. If the philosophy fails my standards, I lose interest quickly.

In that sense, the fantasy was the most difficult for me to distinguish, since both had the same core message. At the end of the day, the AI's passage felt just a bit too long and so I voted the other one.

 

[DismaiNaim]

Is this ai

Your mother is AI

 

[LeilaniOtter]

If you honestly read each of these passages, and I mean really read them, digest them, even picture the words in your head, or the scene being portrayed (except for the science of course), you can choose the human writing every time. AI is going to seem "mechanical" and "direct", with no real flair or creativity. I took the test (the science one got me), and did alright. I spotted the humans.

 

[rileykifer]

That's really interesting. Would you say it was the content of the excerpts or the writing that made you feel bored?

About a mix of both. They all just feel so equally bland to me but I can't figure out why. Maybe it's just not a style I like, or maybe I'd like it more if I read it within the actual book. But I also hate lengthy prose/description, so that could be a big part of it too.

 

[CharlesEBrown]

Just about every passage sounded like AI to me, but my brain is also a bit fried right now. My result ended up being no preference. (Though only three I chose the AI option for, and those were a "this bores me less" type of choice instead of actually liking it.)

Yeah, kind of the same here. I got the first one "right" - but both felt off. Got the science one "wrong" probably because the AI passage reads more like the most recent stuff I read from Carl Sagan (his essay on why we will probably never encounter intelligent life in the stars ... despite him also publishing and co-authoring the novel that was the basis for the movie Contact at roughly the same time the essay came out) than the passage they chose, and I saw that there was a selection from Sagan so was kind of looking for it.
Never read A Wizard of Earthsea - probably should at some point. buy the only Le Guin books I've read were The Left Hand of Darkness and The Beginning Place both for college classes in the late 80s; odds are I would have known her passage if I had; the AI one they had read kind of like something *I* would write so I just sort of gravitated to it.
Really saw no difference in the other ones so just randomly picked one each time and got the "wrong" ones.