171–180 of 233 posts
This article is complete bunk. The researchers used a "chatgpt detector" which as we've seen over and over in academia, do not work. This study is completely unfounded. God I'm choking on the irony of an article about the dangers of using AI to train AI based on a study that used AI to detect AI
Doesn't this play into the whole "Snake eating it's own tail" scenario.. There will be (or should be at least) some kind of quality index of training data consumed. Companies could wear it like a 'quality' badge. Just not sure how you would do it. Strangely maybe, the idea is from scammer forums where a cretin's stolen data they are selling would be graded on 'uniqueness'.
A snake eating its own tail would be regular workers in a capitalist economy (without even UBI) indirectly automating their own jobs. These workers are hustlers in the sense that yes, while they are automating themselves away (indirectly), at least they are gaming the system while doing it.
Earlier quoted context omitted.
I don’t know what a HIT is, but I’m pretty sure the problem is that they wanted you to use your _eyeballs_! Not OCR. And yeah, the service is notorious for underpaying.
Eyeballs are optical, and it sounds like GP was recognizing characters, so I think it was, in fact, Optical Character Recognition.
Earlier quoted context omitted.
I don’t know if it really is a fundamental problem though. Human knowledge was able to bootstrap itself. Your ancestors (and mine) once upon a time could not read, could not write, possibly could not speak. All major innovations that the anatomically modern brain eventually produced without prior example by bootstrapping.
That's because humans don't say things without understanding them. A LLM will parrot what it learned without any understanding, which is unlike a human.
Earlier quoted context omitted.
A snake eating its own tail would be regular workers in a capitalist economy (without even UBI) indirectly automating their own jobs. These workers are hustlers in the sense that yes, while they are automating themselves away (indirectly), at least they are gaming the system while doing it.
This is a lump of labor fallacy. Automation has the opposite effect of what you just said.
(Easy to dismiss things once you give it a name.)
I’m merely taking the LLM disruption hype at face value.
This article is complete bunk. The researchers used a "chatgpt detector" which as we've seen over and over in academia, do not work. This study is completely unfounded. God I'm choking on the irony of an article about the dangers of using AI to train AI based on a study that used AI to detect AI
It's actually pretty easy to create a bespoke ChatGPT detector!
Earlier quoted context omitted.
It's actually pretty easy to create a bespoke ChatGPT detector!
But will it give reliable results?
i think the core problem is with the generalist classifiers (gptzero, openai detector, etc). ex. openai's classifier has an accuracy of around 25% on it's own text. however, when you train a bespoke classifier (like the authors did), you can get really good results.
Earlier quoted context omitted.
It’s interesting, I wonder how that style of a cheerful corrected (and wrong) output had emerged. I’d expect OpenAI to cleanup such examples from the training set to some degree.
I assumed OpenAI had added that themselves.