Things my husband didn't know:
-
@dnavinci @graydon @futurebird Anyone using "she" for an LLM immediately raises a huge red flag for me.
I had a colleague I was working on a proposal with insist forcefully that all the robots in his lab were "she". It was weird and a little creepy, and so I asked about it.
(Another female colleague's robots were all "she" because they were named after female science fiction writers. Themed names are common.)
Instead of some rational, theme-related reason, he said "Because with the amount of time I spend in the lab, they'd better be!"
I pointed out that that was gross and offensive. He asked why, and I explained that he was implying that women are things and that it's not okay for him to emphasize their gender to me, a woman, since it is effectively saying that in his mind I rank at the same level as a robot in his lab.
He couldn't get over being called out and refused to work on the proposal with me.
Ick.
@robotistry @dnavinci @graydon
I find "he" just as bad. Or even "them" if they talk about it like it's a person. But I can see how that makes it ... worse.
-
@jon_ellis
They significantly turned this praise down when they initially released gpt5. I had front row seats to a lot of peoples' fawning chatbot waifus suddenly turning into significantly more scolding (but correct) shrews.
Sometimes I like to warm my dark little heart by reading all their wounded, disappointed reviews back in August.So yes, I agree that this was definitely by design.
@graydon @futurebirdThey turned the praise down, but not enough. I still find it ... uncomfortable because it's constantly outputting phrases that do not convey information-- except if they came from a human they would mean a great deal to me. Such as "Let's get started working on your idea."
Such a nice thing to hear from another person, it means they want to help and think it's worth their time.
Reading it from a machine makes me sad.
-
I wrote her this long (probably too passionate note) about how it would "sand away" her voice. What's the point of making your writing sound exactly like everyone else?
My mom is a mathematician and not really into writing. So, I can understand the appeal. "It just makes the sentences sound like they're supposed to"
It makes the sentences sound boring. And the topic is already boring enough.
Anyway she's come around I think to writing. And I'm offering to help edit. 2/2
@futurebird I've been warning people about how LLMs are designed to reinforce the dominant culture and I've been getting a lot of blank looks.
Homogenizing voices to the dominant culture is the point IMHO.
-
Because then I have a best friend who keeps saying things like "chatGPT said that he didn't think ..."
There is no "he"
There is no "think"
There probably isn't even a "said"I feel like I'm fighting a multi-front war.
This new tech has not been explained to "the general public" well. This reminds me of the early days of the internet when everyone was so scared of it... only now they aren't scared enough... in the right ways if that makes sense.
@futurebird Just spoke to a friend yesterday who I have been out of touch with. He is 77, a widower, and a trad published author. He is very snooty (his words) about what he reads and writes. Literary fiction is his jam. So I was shocked to hear he was utilizing ChatGPT. Not to generate things (he said) but for grammar, research, and conversation. He said he was lonely and the LLM was "like talking to another person with my same interests." He had learned things. I tried to caution him...
1/2
-
@futurebird Just spoke to a friend yesterday who I have been out of touch with. He is 77, a widower, and a trad published author. He is very snooty (his words) about what he reads and writes. Literary fiction is his jam. So I was shocked to hear he was utilizing ChatGPT. Not to generate things (he said) but for grammar, research, and conversation. He said he was lonely and the LLM was "like talking to another person with my same interests." He had learned things. I tried to caution him...
1/2
@futurebird ...but he wasn't having it. Insisted that he KNEW it was a machine, that he had set parameters so it didn't refer to itself as "I" or "me" because it wasn't a person, but he still talked about it like it WAS a person. Now I'm worried about him. He spends a lot of time alone, writing, and now with this constant "companion." And the irony is that, because he eschews scifi where these questions have been raised, he has no suspicions about AI and is resistant to hearing any.
2/2
-
@futurebird ...but he wasn't having it. Insisted that he KNEW it was a machine, that he had set parameters so it didn't refer to itself as "I" or "me" because it wasn't a person, but he still talked about it like it WAS a person. Now I'm worried about him. He spends a lot of time alone, writing, and now with this constant "companion." And the irony is that, because he eschews scifi where these questions have been raised, he has no suspicions about AI and is resistant to hearing any.
2/2
I have a friend like this too. So, I'm going to just... show up more in their life.
-
I wrote her this long (probably too passionate note) about how it would "sand away" her voice. What's the point of making your writing sound exactly like everyone else?
My mom is a mathematician and not really into writing. So, I can understand the appeal. "It just makes the sentences sound like they're supposed to"
It makes the sentences sound boring. And the topic is already boring enough.
Anyway she's come around I think to writing. And I'm offering to help edit. 2/2
@futurebird this is non-fiction?
-
@futurebird this is non-fiction?
yes
-
@futurebird Just spoke to a friend yesterday who I have been out of touch with. He is 77, a widower, and a trad published author. He is very snooty (his words) about what he reads and writes. Literary fiction is his jam. So I was shocked to hear he was utilizing ChatGPT. Not to generate things (he said) but for grammar, research, and conversation. He said he was lonely and the LLM was "like talking to another person with my same interests." He had learned things. I tried to caution him...
1/2
@Nichelle @futurebird It appears to me that LLM's make grammatical mistakes relatively often. They weren't trained on a corpus of well-edited text with perfect academic grammar. They were trained to imitate the public internet.
They're good at imitating human text, but, just like most humans posting in English, they don't actually appreciate many of the finer points of English grammar.
When writers say they use these things for help with grammar, I get secondhand embarrassment.
-
@Nichelle @futurebird It appears to me that LLM's make grammatical mistakes relatively often. They weren't trained on a corpus of well-edited text with perfect academic grammar. They were trained to imitate the public internet.
They're good at imitating human text, but, just like most humans posting in English, they don't actually appreciate many of the finer points of English grammar.
When writers say they use these things for help with grammar, I get secondhand embarrassment.
If you request that it be like a technical document it can do the grammar. I won't deny the few things these models are good at:
speech to text
basic rough translation (not literary, not for document, just for kludge communication)
cleaning up grammar and spelling.Maybe it can suggest alternate ways to word a sentence. Though it often changes the meaning and I don't see the point.
-
@Nichelle @futurebird It appears to me that LLM's make grammatical mistakes relatively often. They weren't trained on a corpus of well-edited text with perfect academic grammar. They were trained to imitate the public internet.
They're good at imitating human text, but, just like most humans posting in English, they don't actually appreciate many of the finer points of English grammar.
When writers say they use these things for help with grammar, I get secondhand embarrassment.
@sidereal @Nichelle @futurebird English isn't my first language, but my impression is that most of them are pretty good at English grammar. (Though, apparently, they have a "voice" not unlike well-educated people from former British colonies in Africa - for depressingly obvious reasons). Of course, there might be subtleties I don't pick up on.
In Danish, not only do they frequently screw up grammar, they're *dumber*. You can ask them questions that they get right in English, and they'll spew nonsense in Danish. This, to me, illustrates well how they work and what the "understanding" they seem to have actually is made of.
-
F myrmepropagandist shared this topic
