#some sunday afternoon girlblogging

[maaarine]

tech people are freaking out about GPT-4 and warning that users can easily be fooled by the chatbot, because it always sounds so autoritative and convincing even when it gives blatantly wrong answers

two thoughts:

1) they don’t seem to worry that this is exactly what Straight White Men sound like all the time

should we also warn and train people not to let their mind be fooled by those men’s air of authority and objectivity? I vote yes

2) it’s interesting that the chatbot’s apparent intelligence is purely based on language

the answers it gives emerges from statistical linguistic probability, from what is most likely to be said in the context

so in the same way that your phone can auto-fill words based on what you usually write, or correct the word “duck” to “fuck” because that’s probably what you really mean, the chatbot also processes the language it’s been fed, and comes up with the most likely sequence of words

and this whole thing reminds me of my niece, who as a very young kid would form complex sentences and sound totally legit, while I knew for a fact that she didn’t and couldn’t understand the meaning of what she was saying

it always struck me, how she managed to sound impossibly smart just by repeating what she’d heard adults say in a similar situation, and mimicking their cadence

the lesson I took from that experience is that this is what happens everywhere all the time, this is how people can sound so convincing without actually understanding anything

there seems to be a fundamental difference and disconnect between verbal ability and... whatever the other thing is — a meaning ability?

like there’s knowledge that exists in language form, and there’s Knowledge that exists in some kind of metaphysical form, and they run on different tracks, and can exist independently of one another

you encounter people who are human versions of the chatbot, who can give you the “right” answers because they’ve absorbed and averaged the language of the context, they’ve memorized the sequences of words from the textbooks

but their verbal ability is not a manifestation of a meaning ability, it’s like being able to solve equations without attaching any meaning to the math

and I don’t mean this in a “those people are sheeple NPCs who can’t think for themselves” kind of way

to be fair, I’m not sure what I mean

just that there are two kinds of “intelligence” at play here in a cognitive sense, and that it’s very hard to delineate and conceptualize them, because the meaning ability exists in an intangible realm

[huellitaa]

☀️🎀 princess project: day 47!

──★ ˙ ̟🎀 sunday 14.7.24

🗒🎀𓂃 ࣪˖ today's to do!

☀️ mental

journalled

been feeling better since my medicine got upped, i think it's helping me !

🎀 physical

drank lots of water ♡

took a cold shower

washed hair ♡

🧁 academic

nothing here... ♡

💬 social

texted my friendssss

chatted w my dad all afternoon

took some time for myself today ♡

🎀 leisure

made my first top!! ♡

binge watched the a good girl's guide to murder series and sobbed violently

remade some jewelry i broke a while ago

worked on my girlblog !!!!

made a pretty charm for my bsf ♡

changed my bed

swept my floors

cleaned down surfaces

tidied up ♡

put clothes away

[maaarine]

when I was a kid, I liked reading sci-fi novels about aliens and space travel, but in a very cool and not at all nerdy way obviously

and more than 20-25 years later, I still occasionally think about one of them

it was the French translation of My Teacher Flunked the Planet (Bruce Coville, 1992), which was the fourth entry in a series called My Teacher Is An Alien

the cover showed an alien about the press a big red button, and lead you to believe that he had pushed it to "flunk” the earth

but the twist of the story, and this is entirely based on my memory of it so I could be misremembering, was that he had indeed destroyed the planet, but not by pushing a button

he had actually contributed to the invention of television, which had inadvertently fucked over culture and people’s mind

obviously this worry about television is dated now, as it was a reflection of the time that you also find in the nonfiction classic Amusing Ourselves to Death (Neil Postman, 1985)

but the reason I still remember that book, despite remembering virtually nothing about my past, is because this idea that destruction would come from a seemingly innocuous piece of communication technology, instead of coming from an obvious weapon, had made a big impression on me at the time

and today when the news jump from talking about Russia’s nuclear power or North Korea launching missiles, to the less concrete concerns regarding GPT-4′s hallucination of false information or Tinder’s gamification of relationships, I still think:

the alien wouldn’t destroy the planet by blowing it up

he would invent GPT-4 and Tinder