Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

That viral Reddit post about food delivery apps was an AI scam

theverge.com

Post reply on HN
Print view View original

41–50 of 55 posts

Re: That viral Reddit post about food delivery apps was an AI scam

#41

Post by waldrews »

stavros wrote: post #39
thot_experiment wrote: post #33

Earlier quoted context omitted.

You know, I wouldn't be surprised if the AI was less than 50% accurate. I'm not claiming that in general, but I'm also certain it would be possible to construct a dataset such that the AI would do far worse than a coin flip.

You know that it's not possible to do worse than a coin flip, right? If you're getting it 100% wrong, I'll just do the opposite of what you say, and have a 100% correct predictor.

The threshold isn't 50% because the distribution of human and AI written cases isn't naturally 50-50. So a coin flip will underperform always guessing the more frequent class. Where it gets interesting is if the base is unknown or variable over time or between application domains. Like, since AI written text is being generated faster than the human kind, soon guessing AI every time will be 99% accurate. That doesn't mean such a detector is useful.

Re: That viral Reddit post about food delivery apps was an AI scam

#42

Post by stavros »

waldrews wrote: post #41
stavros wrote: post #39

Earlier quoted context omitted.

You know that it's not possible to do worse than a coin flip, right? If you're getting it 100% wrong, I'll just do the opposite of what you say, and have a 100% correct predictor.

The threshold isn't 50% because the distribution of human and AI written cases isn't naturally 50-50. So a coin flip will underperform always guessing the more frequent class. Where it gets interesting is if the base is unknown or variable over time or between application domains. Like, since AI written text is being generated faster than the human kind, soon guessing AI every time will be 99% accurate. That doesn't…

When we say "coin flip" in these situations we mean "chance", ie the prior distribution. Otherwise a predictor of the winning lottery numbers that's "no better than a coin flip" would mean it wins the jackpot half the time.

Re: That viral Reddit post about food delivery apps was an AI scam

#43

Post by g8oz »

A side note to this: a lot of the longer highly upvoted Reddit posts in subs like /AskReddit have clearly been written with AI help. The three parallel element structure with a a splashy conclusion sentence for synthesis is a dead giveaway.

I am irrationally irritated by this.

Re: That viral Reddit post about food delivery apps was an AI scam

#44

Post by thot_experiment »

stavros wrote: post #39
thot_experiment wrote: post #33

Earlier quoted context omitted.

You know, I wouldn't be surprised if the AI was less than 50% accurate. I'm not claiming that in general, but I'm also certain it would be possible to construct a dataset such that the AI would do far worse than a coin flip.

You know that it's not possible to do worse than a coin flip, right? If you're getting it 100% wrong, I'll just do the opposite of what you say, and have a 100% correct predictor.

Only if you have that data available to you, the brain to analyze it and the freedom to chose.

Re: That viral Reddit post about food delivery apps was an AI scam

#45

Post by nacozarina »

g8oz wrote: post #43

A side note to this: a lot of the longer highly upvoted Reddit posts in subs like /AskReddit have clearly been written with AI help. The three parallel element structure with a a splashy conclusion sentence for synthesis is a dead giveaway. I am irrationally irritated by this.

that it’s rage-inducing is not yet clearly irrational

Re: That viral Reddit post about food delivery apps was an AI scam

#46

Post by on_the_train »

g8oz wrote: post #43

A side note to this: a lot of the longer highly upvoted Reddit posts in subs like /AskReddit have clearly been written with AI help. The three parallel element structure with a a splashy conclusion sentence for synthesis is a dead giveaway. I am irrationally irritated by this.

It has to be clarified though, that askreddits main purpose is to gather karma for new accounts. It's the only popular sub without karma thresholds. So is common knowledge to spam it with new accounts to make them usable. It's the official advice even I think

Re: That viral Reddit post about food delivery apps was an AI scam

#48

Post by sandyarmstrong »

The Verge's article here is the least compelling. They report it with the same confidence and style as their original article when the Reddit post first made waves.

I recommend reading https://www.platformer.news/fake-uber-eats-whisleblower-hoax... instead. It offers better reasoning, and an honest reflection about how close they were to getting taken in by the original Reddit post.

Re: That viral Reddit post about food delivery apps was an AI scam

#49

Post by waldrews »

stavros wrote: post #42
waldrews wrote: post #41

Earlier quoted context omitted.

The threshold isn't 50% because the distribution of human and AI written cases isn't naturally 50-50. So a coin flip will underperform always guessing the more frequent class. Where it gets interesting is if the base is unknown or variable over time or between application domains. Like, since AI written text is being generated faster than the human kind, soon guessing AI every time will be 99% accurate. That doesn't…

When we say "coin flip" in these situations we mean "chance", ie the prior distribution. Otherwise a predictor of the winning lottery numbers that's "no better than a coin flip" would mean it wins the jackpot half the time.

Yup! My point is that the 'coin flip baseline' model that's as good as chance isn't actually trivial to create, for an unbalanced and time varying underlying distribution.
Post reply on HN