The remarks from the expert in this article, even as critical as they are, grossly understate just how foolish and messed up this is.
-
@scott @inthehands I recently asked a coworker for evidence behind a claim she was making. She said she asked ChatGPT.
@eniatitova As much as that may be foolish without checking ChatGPT's stated source for the information, it's not *as* wrong as using LLMs to replace polling. Information backing up that claim might actually have been in a given LLM's training set. Knowledge of what the whole population thinks about a specific question right now, can't possibly have been.
-
@eniatitova As much as that may be foolish without checking ChatGPT's stated source for the information, it's not *as* wrong as using LLMs to replace polling. Information backing up that claim might actually have been in a given LLM's training set. Knowledge of what the whole population thinks about a specific question right now, can't possibly have been.
@scott I agree, I think people have completely lost the plot on what it is LLMs actually do: generate text. hard stop. I can’t reveal my question because atty client privilege but suffice it to say, there was no source that ChatGPT could have had access to that would have provided a credible evidence.
-
I don’t know, perhaps I read too hastily and am misrepresenting what these people did. But at this point the nonsense is so severe I can believe it without even doing a double take
OK, I looked at it in slightly more detail, and yes, my OP is basically a correct summary of what they’re doing.
Statistical Jesus wept.
-
@inthehands What tripe.
@rjblaskiewicz @inthehands
It's just offal. -
@rjblaskiewicz @inthehands
It's just offal.@gnate @rjblaskiewicz
Ha, nice. I am now going to start using the phrase “offal data” for this kind of thing -
OK, I looked at it in slightly more detail, and yes, my OP is basically a correct summary of what they’re doing.
Statistical Jesus wept.
(The underlying logic here is that LLMs embed biases, so you take advantage of that fact by prompting an LLM to take on a spectrum of different demographic biases that correspond to population demographics, then ask the LLM a polling question in the context of each of those demographically weighted biases.
So yeah, from my OP it might sound like they’re replacing polling with stabbing themselves in the face, but •actually• they’re juggling a bunch of knives and •then• stabbing themselves in the face.)
-
(The underlying logic here is that LLMs embed biases, so you take advantage of that fact by prompting an LLM to take on a spectrum of different demographic biases that correspond to population demographics, then ask the LLM a polling question in the context of each of those demographically weighted biases.
So yeah, from my OP it might sound like they’re replacing polling with stabbing themselves in the face, but •actually• they’re juggling a bunch of knives and •then• stabbing themselves in the face.)
@inthehands
Gah! Wrong damn biases tho.The gd stupidity is just too.much sometimes
-
(The underlying logic here is that LLMs embed biases, so you take advantage of that fact by prompting an LLM to take on a spectrum of different demographic biases that correspond to population demographics, then ask the LLM a polling question in the context of each of those demographically weighted biases.
So yeah, from my OP it might sound like they’re replacing polling with stabbing themselves in the face, but •actually• they’re juggling a bunch of knives and •then• stabbing themselves in the face.)
@inthehands emphasis on the through face stabbing
-
@inthehands emphasis on the through face stabbing
@wronglang
But see they juggled the knives first!They •juggled• them, Krzysztof
-
I don’t know, perhaps I read too hastily and am misrepresenting what these people did. But at this point the nonsense is so severe I can believe it without even doing a double take
@inthehands they were never looking for a survey methodology, they were looking for a beard!
-
@wronglang
But see they juggled the knives first!They •juggled• them, Krzysztof
@inthehands that was so you couldn't tell if they were stabbing their own face or yours
-
RE: https://flipboard.com/@futurism/futurism-1lupih3cz/-/a-hWRkyR9zQFuooBpWKGziAw%3Aa%3A1737388686-%2F0
The remarks from the expert in this article, even as critical as they are, grossly understate just how foolish and messed up this is.
“We didn’t conduct a poll, we just asked GPT or whatever to make up things people would say and then sampled that”
My dudes. No.
@inthehands FFS
-
RE: https://flipboard.com/@futurism/futurism-1lupih3cz/-/a-hWRkyR9zQFuooBpWKGziAw%3Aa%3A1737388686-%2F0
The remarks from the expert in this article, even as critical as they are, grossly understate just how foolish and messed up this is.
“We didn’t conduct a poll, we just asked GPT or whatever to make up things people would say and then sampled that”
My dudes. No.
Next it will be asking LLMs what voters would choose and just going with that. No need for elections.
-
RE: https://flipboard.com/@futurism/futurism-1lupih3cz/-/a-hWRkyR9zQFuooBpWKGziAw%3Aa%3A1737388686-%2F0
The remarks from the expert in this article, even as critical as they are, grossly understate just how foolish and messed up this is.
“We didn’t conduct a poll, we just asked GPT or whatever to make up things people would say and then sampled that”
My dudes. No.
@inthehands @mastodonmigration @Futurism
I think polls should all include the question:
Did you find any of these questions leading?
And be required to publish that result next to any results along with those that commissioned, paid for, the poll.
-
Say what you will about predicting the future from animal entrails, but at least that can give you dinner
@inthehands Can I quote you on that?
-
@inthehands Can I quote you on that?
@drwho
Any time -
@drwho
Any time@inthehands Thank you kindly.
-
RE: https://flipboard.com/@futurism/futurism-1lupih3cz/-/a-hWRkyR9zQFuooBpWKGziAw%3Aa%3A1737388686-%2F0
The remarks from the expert in this article, even as critical as they are, grossly understate just how foolish and messed up this is.
“We didn’t conduct a poll, we just asked GPT or whatever to make up things people would say and then sampled that”
My dudes. No.
@inthehands This is just a fancy new way of making shit up.
Publishing fabricated polls as if they were legit, is called "fraud".
-
OK, I looked at it in slightly more detail, and yes, my OP is basically a correct summary of what they’re doing.
Statistical Jesus wept.
@inthehands And not just statistics! I’ve also seen certain linguists and philosophers excited about the possibility of posing grammatical/ethical judgments to LLMs in bulk, apparently fully convinced that the results would hold any meaning or value whatsoever. Can’t wait for the smug “Trolley problem objectively solved” papers
-
OK, I looked at it in slightly more detail, and yes, my OP is basically a correct summary of what they’re doing.
Statistical Jesus wept.
@inthehands 97.5% of Jesuses wept, +/- 2%.
