Conversation

Notices

1. Embed this notice
   Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 29-Aug-2025 07:43:34 JST Paul Cantrell

   I suspect we’re doing something similar to the tank classifier when we ascribe intelligence to AIs: there are patterns such as (for example) grammatical correctness which we •associate• with this abstract thing called “intelligence,” and we thus mis-infer the existence of everything else we associate with the notion of “intelligence” when we see (for example) correct grammar.

   Or should we just call the machine intelligent because we classify it as intelligence because our brains, which we assume are intelligent, think it fits the pattern of intelligence? And now you see what the OP means about “begging the question of intelligence.”

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 29-Aug-2025 10:19:05 JST Paul Cantrell

     in reply to

     • Wayne Werner

     @ketmorco
     Ha, that’s good. (I mean, for some values of “works” it totally does work — but if we’re talking about it being “intelligent” or “close to AGI” or whatever, then what you said)

   • Embed this notice
     Wayne Werner (ketmorco@fosstodon.org)'s status on Friday, 29-Aug-2025 10:19:06 JST Wayne Werner

     in reply to

     @inthehands I like the explanation that AI doesn't work because it's intelligent, it works because we see Jesus in toast

   • Embed this notice
     Wayne Werner (ketmorco@fosstodon.org)'s status on Friday, 29-Aug-2025 10:19:07 JST Wayne Werner

     in reply to

     @inthehands I like machine learning because it's what it is. The machine can learn but it's still just a sand+electricity machine and it's learning isn't actually intelligence by any stretch of the word.

     It doesn't derive axioms of the system that it's in or built up of, and wouldn't even be able to understand or argue about *that*.

     We just think it can because it presents words in the kind of order that we think it should present. It's really just a fancy way of making https://www.tylervigen.com/spurious-correlations

   • Embed this notice
     Carsten Franke (carstenfranke@mastodon.social)'s status on Friday, 29-Aug-2025 10:21:25 JST Carsten Franke

     in reply to

     @inthehands what would happen if we called it simulated intelligence, SI? An artificial object implies that it still has the properties of the original, while a simulated object only pretends to be the same as the original.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 29-Aug-2025 10:21:25 JST Paul Cantrell

     in reply to

     • Carsten Franke

     @carstenfranke “Simulated” is good, though it’s ”intelligence” that’s the real tarpit here

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 29-Aug-2025 10:26:51 JST Paul Cantrell

     in reply to

     • Greg Bell

     @ferrix
     I am grudgingly inured to the latter definition being at this point widely enough used to have become correct. Alas.

   • Embed this notice
     Greg Bell (ferrix@mastodon.online)'s status on Friday, 29-Aug-2025 10:26:52 JST Greg Bell

     in reply to

     @inthehands I approve of calling out cases of "begging the question", both in reasoning and when mis-used to mean "leading to the question"