Jon Stokes (@jon_stokes) header
Jon Stokes (@jon_stokes) profile picture

Author. Coder. CTO. θηριομάχης. Building: https://teapaw.org/t.co/otXT4Wy6WR. Writing: https://teapaw.org/t.co/dBPBtyCIHw.

Austin, TX
Joined July 2008
Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)Photo by Jon Stokes (@jon_stokes)
All the blackpilling on my TL has me worried some of you will turn into preppers, so here's a get-it-over-with list by a former hardcore prepper that'll let you scratch that itch then move on with your life.
74
134
68
1,948
601,908
If you are going to send me a cold email then please use enough AI to avoid stuff like this
1
359
Jon Stokes retweeted
Nagel isn’t making some dumb mistake. No one seems to understand what “natural manifestation” means. He likely means two things, both of which are coherent. (1) consciousness comes before language, and language is an abstraction of it. (2) there is no such thing as a “purely linguistic experience” for humans, in that language does not naturally manifest in consciousness. You are conscious of language as heard words (LLMs can’t hear) or as it relates to the actions and dispositions of your embodied self (they aren’t embodied either).
This argument from Thomas Nagel is... terrible. Just bizarre. It seems pretty unambiguously to beg the question. Odd to see such a prominent philosopher make such an obvious mistake.
31
27
5
284
16,175
Jon Stokes retweeted
What they want - OAI: capture light cone of all future value. Anthropic: rule world because no-one else is worthy. Google: meet 4th quarter analysts' expectations. ...and that's why I want Google to win.
61
91
10
1,502
36,127
Jon Stokes retweeted
Ah yes, Lord of the Rings, the famous Tolkien parable about how we should totally just grab the ring first!
19
56
20
669
100,006
Jon Stokes retweeted
The T. Rex Arms Field Manual Vault just got upgraded with its "Download All" button. And it's not just 4,200 manuals, it is now 6,400. And it's not just PDF and ePub, now it includes MD files. And it's not just streaming from our site, it's on the torrent networks.
87
169
25
1,971
53,597
Jon Stokes retweeted
Just quoting Thomas Nagel: “There’s no mental life there in the ordinary sense. It’s not a mind. It’s not somebody that we have to care about. It’s a machine. Here are some of the smartest people on the planet being completely idiotic”
15
32
3
192
7,087
Jon Stokes retweeted
Replying to @Dr_Atoosa
One prediction I feel fairly confident about is that clips from such videos will appear in future documentaries, showing how some very smart people circa 2026 made preposterously confident proclamations about what was to come
1
2
67
2,507
Jon Stokes retweeted
I’m personally quite tired of these pretty much unbased universalist claims made by people with perhaps zero to super minimal knowledge of social sciences, history, and humanities. There are many forms of market organisation and employment, both existing and possible, that I’m almost sure you haven’t even considered. The public need to think much more maturely about society and economics and, ahem, show a little more epistemic humility when making claims about “all humans.”
Anthropic Researcher Tao Lin: Pretty much all humans on earth will become permanently unemployable because of AI "It's very likely that pretty much all humans on earth will become permanently unemployable" "And will need to strictly rely on government welfare to sustain themselves" "Human employment will go away."
58
115
5
1,134
47,781
Believe in myself? The same person who got me into this shit?
47
3,377
86
21,529
264,322
Jon Stokes retweeted
p(Doom) combines the false precision of a numeral with the handwaving vagueness of apocalyptic scenario-making. @TedNordhaus and me on how to improve the AI safety conversation. breakthroughjournal.org/p/pd…
6
1
17
2,058
Jon Stokes retweeted
These kinds of dramatic proclamations are not credible. Obviously we don’t know that a policy of moderate AI use at universities will lead to the collapse of mass literacy.
Always worth reiterating re higher ed AI policy that the choice is not between extreme anti-AI measures and moderation. There are only two choices, both extreme: teach like it’s the 1990s, or accept the slow but steady collapse of liberal education and mass literacy.
2
1
25
2,120
Jon Stokes retweeted
People keep asking me about ham radio for prepping, and I’ll give my honest answer: It’s not worth it. You’re better off with GMRS for short-distance comms and a Starlink Mini for SHTF internet connectivity. I have a General ham license (KO4EMJ) but have never once used it for emergency comms.
2
1
16
1,156
Jon Stokes retweeted
Joe got shit for this but the Pangram discourse is a point for him. The stochastic parrots people have gone from anger (AI is worthless) to denial (AI detectors don't work); next is bargaining (objecting to AI prose is elitist) and acceptance (AI is a vital accessibility tool)
Because the left assigns a lot weight to the needs of immigrants, ESL speakers, people with mental health issues, people in unconventional romantic relationships and so forth and all of these people would suffer if AI were decommercialized .
10
16
3
296
37,593
Jon Stokes retweeted
If the EAs manage to get a bunch of extremely intelligent, data-facile 30-somethings to become line Hill staff, they’ll utterly rule the entire Congress within three years. It’ll be like those historical fiction books asking “what if the Roman Empire had tanks?” Just domination.
ultimately it's just very difficult to beat a philosophical/ideological/movement that can inspire Jane Street traders or Google engineers to quit their high-paying jobs in California to work with a bunch of relatively clownish coworkers on the Hill for $60k/yr in 84° DC summers
8
8
9
197
24,346
This is the key point
Some quick reactions to the recent Dwarkesh podcast with Paul Christiano. Tl,dr: bullish on rapid capability gains, more skeptical of the leap to takeover, and wary of emergency overreach.
4
10
1
46
3,543
No you're right. I made this argument back in March & I'm even more convinced of it now that public data shows most of the lab spend going to post-training.
Am I wrong in thinking that the massive discrepancy here reveals that some huge amount of frontier model capability right now must be from high-effort, bespoke data collection rather than "runaway generalization"?
6
17
1
241
11,995
To the bio point, Anthropic is quite openly making a big bio push (which also makes sense given Dario's background), so they're training on that data and yeah Claude is getting good at it.
2
8
1,056
Cool chart from @SemiAnalysis_ showing that the frontier labs have now shifted most of their compute from pre-training to post-training (mostly RL). Unlike pre-training, the bottleneck moves from compute to networking - how fast you read from memory/access CPU. An interesting implication is that you don't need to stack memory (DRAM dies) anymore as this doesn't improve bandwidth so we've gone from 12 dies ---> 8 dies ---> 4 dies? So, I'm guessing companies that do scale up and scale in win here?
3
800
Have you seen the new supply chain attack? It's on Optimizely. It's on Bitly. You can literally experience it on United. It's on Bank of America with ads. Dude it's on Chase. It's an FAA original. It's on KLM. You can go to CNN and experience it
JUST IN - A suspected supply chain attack on Optimizely loaded malware on Zoom, CNN, Fox News, BBC, Salesforce, GitLab, KLM, Air France, Bitly, IG Group, American Eagle and others, exposing visitors to scripts built to steal payment data and exploit iOS devices.
15
448
11
5,829
97,366