First of all, precisely because it is LLM, it is good to point out that it is a source.
Second, the LLM and chat GPT4 in particular can often summarize information quite well. Yes, there are cases when it hallucinate, but at this point it is very rare for GPT4. So, when I do not have hours and hours to spend on some topic like this, asking ChatGPT4 is a valid strategy to get reasonable probability correct answer, with much higher probability than random claims here in fediverse.
So, I will defend the method as reasonable, and will ask you, do you claim that what is stated is false?
You are posting a position you don’t even necessarily understand. You are providing what might be facts or maybe not. Might be subject to bias in training data. What do you think this adds to the discussion? You want someone to disagree with stuff you don’t even know?
I don’t have time for that bullshit so I asked ChatGPT to write a rebuttal. I’m not even going to post the whole thing, just the following excerpt:
“It’s crucial to understand that transgender women are individuals whose gender identity is female, regardless of their assigned sex at birth.”
This right here is why posting AI answers is garbage. It’s not just flat out wrong, but insists that this nonsensical garbage is crucial to your understanding.
And for what it’s worth I’m a big fan of AI. I use it all the time. I’ve written applications that leverage it. I’ve had it help me with coding issues. But I never, ever trust it.
I am not saying ChatGPT4 is fool proof, but neither is any source. And if I actually will try to understand the original sociological research paper, it may increase my chances of misunderstanding the data.
Let me put this this way - start asking ChatGPT4 (not 3.5) biological and medical questions and keep tally of correct answers. You would see how accurate it is. I would say with 95% probability it would answer such questions correctly. And this is how much credence I put in its answer.
So, when you say that I was providing the fact that may be right maybe not, yes, that’s correct. But it is not 50/50, far from it.
And if you do not want to provide rebuttal, it is your right, of course. But then, what’s the point of your post? We just need to believe your statement when you itself refuse to provide zero evidence?? You understand how it looks, right?
In the short term, yes. What this leaves out is that two years of HRT is enough to negate those physical benefits. Hormones are powerful shit.
Also, no it fucking isn’t a reasonable method. It has neither the credentials to know what it’s talking about, nor any obligation to verify that what it says is true. Imagine reading 10,000 shitty sci-fi novels and 1 textbook and thinking you can piece together advanced physics. It literally cannot tell the difference between fact and fiction, nor does it care. It’s a machine. Garbage in, garbage out.
I suggest you to ask ChatGPT4 questions in medicine and biology, and keep tally. If you truly think ChatGPT4 is garbage, you will be surprised. I spent lots of hours interacting with it, and I understand its limitations and strengths. And these kind of questions it usually answers quite well.
I spent lots of hours interacting with it, and I understand its limitations and strengths.
Considering you think it’s a substitute for a scholarly source, I doubt that. Once again, this is a machine designed to repeat things it heard. It’s a mechanical parrot. ChatGPT4 did not earn a degree. It did not study. It does not fact check. It does not give a solitary fuck about the scientific method. If you cannot see why this would be a problem for its credibility, then I can’t help you.
The rest of it
You just tried to use a glorified markov chain in an argument. Suffice to say I do not believe that you are the best judge of factual accuracy in regards to said tally.
I suggest you yourself test it, do not rely on me. It is experimentally can be shown as quite good predictor for these kind of questions. Don’t want to test yourself and don’t believe me? There a lot of tests were done of these models showing that they are already at the level to pass many exams. Your claim that it does not have any credibility is totally unfounded.
Also, I never claimed that it is a substitute for scholarly source, I would never use it in a scientific paper. But I would not use Wikipedia either. But we are on internet on discussion board, the standards here are different. At least I supplied a source, majority of posts here don’t do that, including your statements, by the way, implying that ChatGPT4 has no credibility.
Even if we assumed ChatGPT were completely accurate, that answer includes enough weasel words so it is not sufficient here.
varies significantly among individuals … might retain …. may not completely reverse
All this tells us is that making this fair is not an easy answer, anything more is individual interpretation. It probably needs some sort of medical consensus.
How did they choose four years? Did they pull that out of their asses, or was there medical input? Is there a reason to expect ChatGPT to be a better source than what they used?
Four years was number given in the question, because this number was selected by boxing rules.
As for the rest, are you suggesting that a post for discussion on internet should have the same level of credence as scientific article or article in encyclopedia? Why suddenly such super-high standards to my poste? Coincidently, no-one, despite criticism of my use of ChatGPT, pointed on even single mistake in that text.
Others pointed to it as being non-authoritative: we don’t need to prove it wrong, just that it doesn’t know. This is like listening to Aunt Marge on Facebook: I don’t have to know whether she’s right to be skeptical of her as an “expert”.
my contention was that even if we assume Au t Marge is right, the answer is insufficient fr the question.
You’re complaining that four years of hormone therapy is insufficient without knowing where that came from or why, but we’re saying we’re not listening to Aunt Marge as more of an authority.
I have no idea whether their decision is sound or what facts they base it on, but I’m also not taking Au t Marge’s word for it that they’re wrong
Imagine using predictive text as a source.
First of all, precisely because it is LLM, it is good to point out that it is a source.
Second, the LLM and chat GPT4 in particular can often summarize information quite well. Yes, there are cases when it hallucinate, but at this point it is very rare for GPT4. So, when I do not have hours and hours to spend on some topic like this, asking ChatGPT4 is a valid strategy to get reasonable probability correct answer, with much higher probability than random claims here in fediverse.
So, I will defend the method as reasonable, and will ask you, do you claim that what is stated is false?
You are posting a position you don’t even necessarily understand. You are providing what might be facts or maybe not. Might be subject to bias in training data. What do you think this adds to the discussion? You want someone to disagree with stuff you don’t even know?
I don’t have time for that bullshit so I asked ChatGPT to write a rebuttal. I’m not even going to post the whole thing, just the following excerpt:
“It’s crucial to understand that transgender women are individuals whose gender identity is female, regardless of their assigned sex at birth.”
This right here is why posting AI answers is garbage. It’s not just flat out wrong, but insists that this nonsensical garbage is crucial to your understanding.
And for what it’s worth I’m a big fan of AI. I use it all the time. I’ve written applications that leverage it. I’ve had it help me with coding issues. But I never, ever trust it.
I am not saying ChatGPT4 is fool proof, but neither is any source. And if I actually will try to understand the original sociological research paper, it may increase my chances of misunderstanding the data.
Let me put this this way - start asking ChatGPT4 (not 3.5) biological and medical questions and keep tally of correct answers. You would see how accurate it is. I would say with 95% probability it would answer such questions correctly. And this is how much credence I put in its answer.
So, when you say that I was providing the fact that may be right maybe not, yes, that’s correct. But it is not 50/50, far from it.
And if you do not want to provide rebuttal, it is your right, of course. But then, what’s the point of your post? We just need to believe your statement when you itself refuse to provide zero evidence?? You understand how it looks, right?
Fool proof*…
Thanks, corrected.
You aren’t correcting anything. You are using chatgpt as a source and using adjectives you don’t even understand. You failed the test.
Its actually foolproof… Well in your case, maybe not so much.
In the short term, yes. What this leaves out is that two years of HRT is enough to negate those physical benefits. Hormones are powerful shit.
Also, no it fucking isn’t a reasonable method. It has neither the credentials to know what it’s talking about, nor any obligation to verify that what it says is true. Imagine reading 10,000 shitty sci-fi novels and 1 textbook and thinking you can piece together advanced physics. It literally cannot tell the difference between fact and fiction, nor does it care. It’s a machine. Garbage in, garbage out.
I suggest you to ask ChatGPT4 questions in medicine and biology, and keep tally. If you truly think ChatGPT4 is garbage, you will be surprised. I spent lots of hours interacting with it, and I understand its limitations and strengths. And these kind of questions it usually answers quite well.
Considering you think it’s a substitute for a scholarly source, I doubt that. Once again, this is a machine designed to repeat things it heard. It’s a mechanical parrot. ChatGPT4 did not earn a degree. It did not study. It does not fact check. It does not give a solitary fuck about the scientific method. If you cannot see why this would be a problem for its credibility, then I can’t help you.
You just tried to use a glorified markov chain in an argument. Suffice to say I do not believe that you are the best judge of factual accuracy in regards to said tally.
I suggest you yourself test it, do not rely on me. It is experimentally can be shown as quite good predictor for these kind of questions. Don’t want to test yourself and don’t believe me? There a lot of tests were done of these models showing that they are already at the level to pass many exams. Your claim that it does not have any credibility is totally unfounded.
Also, I never claimed that it is a substitute for scholarly source, I would never use it in a scientific paper. But I would not use Wikipedia either. But we are on internet on discussion board, the standards here are different. At least I supplied a source, majority of posts here don’t do that, including your statements, by the way, implying that ChatGPT4 has no credibility.
Even if we assumed ChatGPT were completely accurate, that answer includes enough weasel words so it is not sufficient here.
varies significantly among individuals … might retain …. may not completely reverse
All this tells us is that making this fair is not an easy answer, anything more is individual interpretation. It probably needs some sort of medical consensus.
How did they choose four years? Did they pull that out of their asses, or was there medical input? Is there a reason to expect ChatGPT to be a better source than what they used?
Four years was number given in the question, because this number was selected by boxing rules.
As for the rest, are you suggesting that a post for discussion on internet should have the same level of credence as scientific article or article in encyclopedia? Why suddenly such super-high standards to my poste? Coincidently, no-one, despite criticism of my use of ChatGPT, pointed on even single mistake in that text.
It doesn’t need a mistake.
You’re complaining that four years of hormone therapy is insufficient without knowing where that came from or why, but we’re saying we’re not listening to Aunt Marge as more of an authority.
I have no idea whether their decision is sound or what facts they base it on, but I’m also not taking Au t Marge’s word for it that they’re wrong