Stephan Hurtubise

1.6K posts

Stephan Hurtubise banner
Stephan Hurtubise

Stephan Hurtubise

@GravityAllen

Linguist, in principle! Parameters may vary. Also writer, prompt engineer (it’s a thing), and director of @TheLingSpace.

Montreal Katılım Haziran 2010
195 Takip Edilen61 Takipçiler
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@StatedClearly And so we can’t ever truly know what’s been selected for (and why). I’m probably not fully doing the objection justice. But, then, I’m not sure how much justice can be done either.
English
1
0
1
43
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@StatedClearly This is inspired by (somewhat silly, in my view) philosophical objections I’ve encountered w.r.t. whether evolution has explanatory power. The objection. basically runs that traits are complex and rarely unitary – they come bundled with other properties.
English
1
0
0
19
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@StatedClearly Which is a little jarring for a linguist, admittedly. Linguists do talk about content and information and the like, but don’t tend to think about it like an information theorist, since there’s no big incentive to pin it down and quantify it. At least, not from what I’ve seen.
English
0
0
0
13
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@StatedClearly The “bit” in the paper you mentioned isn’t from linguistics, but information theory. It’s a measure of “surprisal” – how expected or unexpected a given syllable is, relative to the syllables that came before; the more predictable (the less surprising), the less informative it is.
English
1
0
1
19
Stated Clearly - Jon Perry
Stated Clearly - Jon Perry@StatedClearly·
If we cracked the mechanics of conciousness and built a machine that experiences constant, maximum fulfillment, would we be morally obligated to build more of them?
English
2
0
0
194
Stated Clearly - Jon Perry
Stated Clearly - Jon Perry@StatedClearly·
I just want to say how happy I am that Nancy @SciBugs found a neon pink Jesus bus for this year's Evolution Tour in the Amazon Jungle. I'm so happy this thing unironically exists!
Stated Clearly - Jon Perry tweet media
English
1
0
2
143
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@MasterTimBlais If you ever follow that path, lemme know how it turns out. :-( I might need to take it seriously.
English
0
0
0
31
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@StatedClearly For instance, applying formal language theory to the study of its nature and level of complexity. Anyway! Great video as always! Congratulations on the milestone!
English
0
0
1
20
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@StatedClearly As in they emerge naturally, without our intervention (and in spite of it, in fact). All evidence points to our having nothing whatsoever to do with the structure of language. So we really shouldn’t be afraid of that category, either – especially if it’s fruitful.
English
1
0
1
19
Stated Clearly - Jon Perry
Stated Clearly - Jon Perry@StatedClearly·
With the exception of several really good research projects on the evolution of de-novo genes, every Templeton paper I read reminds of conversations I had in high school with kids from the "Christian stoner" clique. Am I the only one noticing this?
English
1
0
0
157
Stephan Hurtubise
Stephan Hurtubise@GravityAllen·
@langofmind Thanks for the nice rundown! A good resource, given I’ve been having a lot of similar discussions.
English
0
0
1
43
Ryan Rhodes ⚙️🧠
Ryan Rhodes ⚙️🧠@langofmind·
Chomsky's linguistic claims have not been disconfirmed by LLMs. He has made two principle claims that some have argued LLMs disprove: that it's impossible to acquire a grammar from limited input alone (PoS), and that LLMs are unconstrained and can learn impossible languages
Aryeh Kontorovich@aryehazan

Chomsky did some good math (look up Chomsky–Schützenberger) his linguistics is elegant but made empirically testable claims that are basically being disconfirmed by LLMs as for his politics, the less said the better is he still alive btw

English
8
12
133
10.6K