Founder of Silver Bulletin: natesilver.substack.com/, and cohost of Still Counting podcast!
New York
- Tweets 45,270
- Following 1,760
- Followers 3,042,418
- Likes 10,621
Anyway, today's update is out, fairly boring day of polling for a change.
natesilver.net/p/nate-silver…
2
1
22
24,548
The Vermont governor race is a fun one. Only UNH has polled the race, and they now have the Democrat ahead. But Scott won by 52 points in 2024 and all the expert raters have it solid R! FLIPR has no idea.
18
17
434
58,259
I wrote a whole book on this subject, and the thesis is about the normative and practical importance of human agency. It doesn't help that people like you have no idea WTF AI can do right now.
There's no word for the actual problem going on with AI discourse, which is not that we're anthropomorphizing AI systems but that writers simply airbrush out of the narrative the humans building and running these systems.
23
4
212
105,566
The whole problem is that the AP's language is inaccurate and politically charged. I don't particularly trust the labs. But most (certainly not all) people with real expertise in this field would consider their guidance to be presumptive at best.
You’re policing language right now. You’re policing the AP’s language. Attempting to find more accurate and less industry propaganda terms isn’t policing anything, they’re not imposing their guidelines on Nate Silver, you’re welcome to parrot EA cultists all you’d like
22
4
153
99,893
I think that's an OK description of LLMs from 3 years ago, but a poor one now. A lot of this changed with the introduction of the reasoning models. I don't even think Sam/Dario know *exactly* what their products do but it's a lot more robust than "predicting the next token."
69
3
126
74,541
AI "chatbots" are at most a marginally useful technology. It's the programming and math stuff where they cross into the realm of thinking ("thinking"). Critics without programming/math/tech-adjacent skills are like blind men trying to understand an elephant.
37
7
230
43,343
OK then give Astra or Fable a 10,000-line program, ask it to debug it, ask it to understand what's going on, ask it to make revisions. It's certainly not perfect. I'm bullish on human expertise fwiw. But it sure as hell is way more than a "stochastic parrot."
No, it's different in this case, because LLM are designed to create an illusion that they do think and talk like human beings.
146
27
1,038
260,350
I don't know Eoin. But at the risk of stereotyping based on his tweets, I imagine he's never used AI for programming or anything that really shows off its capabilities. If I were just a Take Artist instead of also having some programming acumen, I'd be way more skeptical.
12
4
69
30,161
IDK if you polled some unbiased sample of experts in the field (granted, "experts" and "unbiased" are doing a lot of work there), I imagine most of them would roughly agree with the idea that AI models "think," at least for more advanced tasks.
60
5
220
127,425
Yeah, there's some threshold of emergent complexity beyond which there's no appropriate substitute for "think". "The computer thinks 2+2 = 4" is weird and wrong, but "Claude thinks" or "Astra thinks" seems mostly fine to me. "Feel" is much more epistemically complicated.
17
80
5,375
Anyway, the continued gains on the generic ballot (and continued decline in Trump approval) has been enough to offset the NYT poll in Maine in our forecast update today: natesilver.net/p/nate-silver…
2
7
71
32,059
Interesting that Dems' generic ballot gains have come from undecideds rather than Republicans, whose numbers have held steady.
58
79
1,134
103,421
An Orca cunningly breaking out of captivity at Sea World, orcathromorphizing is still allowed per @AP.
4
2
70
9,547
I'm mostly talking about posters, politicians are in a different category because they potentially face consequences for their takes. Like I think I have a pretty good mental model for who is going to be Zitron-pilled.
6
1
20
7,250
I mean on the left I think it kind of does map to the Bernie left versus the woke left, the former are much more concerned about AI safety.
9
1
38
16,913
The whole "don't anthropomorphize AI" thing is very Woke 1.0. People trying to police language that is perfectly natural in some cases and ending up in a web of contradictions.
I don't think this is analytically helpful. But either way, extremely surprised to see the line "such as animals" in this statement.
245
47
1,035
389,882
Nate Silver retweeted
This is interesting, via @NateSilver538:
In 2024, one takeaway was that the Red Wave was especially strong in blue cities. In 2026, the Blue Wave seems especially strong in red states.
19
67
451
70,342
I went into that a bit here, as well as in this week's Still Counting. This probably makes for more of a boom-or-bust story than if the gains were concentrated more evenly.
natesilver.net/p/the-blue-wa…
6
10
157
77,787