Matthew Hutson@SilverJacket

Freelance science writer for The New Yorker, Science, Nature, etc. Fire dancer. Into cognition—animal and mineral (aka psych & AI).

 NYC
MagicalThinkingBook.com
Joined March 2009
161 Photos and videos
AI might find new physics where we haven’t looked. My latest for @IEEESpectrum: spectrum.ieee.org/particle-p…

Can AI Find Physics Beyond the Standard Model?

Particle physicists are using machine learning to sift billions of collisions, looking for rare signals of new physics.

spectrum.ieee.org
2
2
12
2,260
Replying to @billwasik
@billwasik Hi, Bill. Great meeting you Thursday at @Caroline_Webb_'s book party. Curious if you found any emails I sent you (from matt@silverjacket.com) in February, August, or September.
1
1
207
@billwasik Thanks for emailing back! Did you see my replies? What do you think of the Causal Labs story?
38
Bland New World: Is AI Making Us All Think the Same? My latest for @Nature Thanks, @ZSourati, @em_wenger, @morwork, @acerbialberto, Alwin de Rooij nature.com/articles/d41586-0…

Bland new world: is AI making us all think the same?

Nature - Researchers worry that generative AI is homogenizing culture and cognition.

nature.com
1
1
227
Researchers worry that AI is homogenizing cognition and culture. We may fall prey to groupthink, @em_wenger warns: “We have suffered as a society when we silence or ignore edge voices.”
1
74
Another method to increase diversity in LLM responses:
Carrie Yuan 袁嘉仪@carrieyuanjiayi
Sep 15
🧬 "In the course of evolution, nature has gone to endless trouble to see that every individual is unlike every other individual." Yet LLMs are forcing us all to be alike by giving us mode-collapsed, homogeneous responses. We propose a multi-agent RL method to fix this. 🧵
60
Headline from @9to5mac: "Samsung reportedly rushing new trifold phone in response to iPhone Duo". Vibes of: theonion.com/fuck-everything…

Fuck Everything, We're Doing Five Blades

Would someone tell me how this happened? We were the fucking vanguard of shaving in this country. The Gillette Mach3 was the razor to own. Then the other guy came out with a three-blade razor. Were...

theonion.com
1
186
One of the most heartwarming messages I saw at Burning Man this year was on a big flag on a man's bike: "SPREAD CHEEKS NOT HATE” I mean, who doesn’t love to smile? :-)
215
Applies to more than just DJing.
ꙮ ts annie 🇵🇸@acephallus
Aug 23
this is who you do it for, not the booth baddies, not the coke guys, but Him
1
311
“Inventing Eliza” Review: The Mother of All Chatbots. My latest book review for @WSJ @WSJBooks. Book by @SarahCiston, @davidmedberry, Anthony C. Hay, @markcmarino, Peter Millican, Arthur I. Schwarz, Jeff Shrager, and @peggyweil. wsj.com/arts-culture/books/i…

‘Inventing Eliza’ Review: The Mother of All Chatbots

Eliza’s creator aimed to fool people into seeing a machine as humanlike then show them how they’d been fooled. The illusion worked too well.

wsj.com
2
1
2
227
The 1960s chatbot, the authors write, “embodies a particular historical moment in both computing and psychology, reveals sophisticated understanding of human conversation, and demonstrates how even primitive AI can raise profound questions about understanding and intelligence.”
1
73
Matthew Hutson retweeted
Arnold Schwarzenegger
@Schwarzenegger
Aug 19
Replying to @adcock_brett
You should melt them.
415
1,750
28,472
946,734
How imminent is recursive self-improvement (RSI)? Researchers asked AI to conduct AI research and write two papers. The results left much to be desired. My latest for @Nature. Thanks, @sayashk and @cong_ml. nature.com/articles/d41586-0…

AI isn’t ready to research itself

Nature - An agentic system successfully developed concepts from two computer-science papers — but the original authors were not impressed.

nature.com
1
5
192
When people aren’t skilled enough to know how unskilled they are: I discussed the Dunning-Kruger effect with @daviddunning6 for @NautilusMag: nautil.us/why-we-think-we-kn…

Why We Think We Know More Than We Do

Why We Think We Know More Than We Do: Breaking down the famous “Dunning-Kruger effect” with David Dunning himself

nautil.us
1
1
658
Was this @nytimes metaphor written by AI? "As with a mail distribution hub, putting computing power closer to users enables A.I. models to think longer, respond more quickly and complete more complex tasks." Data-center proximity doesn't affect a model's ability to think longer or do more complex tasks. And what does it mean for a mail hub to think longer? nytimes.com/interactive/2026…

The Impending, Inescapable Deluge of A.I.

The number of A.I. chips that provide the computing power to advance the fast-evolving technology is doubling every nine months.

nytimes.com
2
6
379
The World Cup final crowd cam cut from Will Ferrell to Pharrell Williams. A+
1
213
Can LLMs reach "AGI," or do we need world models? My latest news feature for @ScienceMagazine (link in reply). Thanks to Jared Kaplan, @ylecun, @PimDeWitte, @AdamJelley2, @janexwang, @shlomifruchter, @LakeBrenden, @yukez, @merettm, @AnthropicAI, @amilabs, @gen_intuition, @GoogleDeepMind, @OpenAI, etc.
1
8
24
11,526
science.org/content/article/…

As better chatbots get harder to build, AI turns to simulated worlds

In pursuit of human-level intelligence, researchers are developing agents that learn by acting in virtual environments rather than simply absorbing more text

science.org
2
314
Matthew Hutson retweeted
Mauro Osorio ⭐⭐⭐@mausorio
Jun 17
Me emociona que ni camine ya, paradito nomás.
LSPN_FC
857
6,843
140,581
3,978,024
Matthew Hutson retweeted
Formula 1
@F1
Jun 14
The moment of victory! #F1 #BarcelonaGP @LewisHamilton
575
18,851
84,558
5,057,105
There’s no way they won’t notice
돌아온평직 (돌평)
147
1,801
17,507
762,659
IPO for @SpaceX
90
Load more