Steve Huntsman
@stevehuntsman.bsky.social
American with bittersweet memories of living in a shining city on a hill. Mathematician. Give em hell Devils. Go Navy. русский военный корабль, иди на хуй. Currently clean on OPSEC. Entropy always wins.
Pinned
Steve Huntsman
@stevehuntsman.bsky.social
· Mar 22
Kid had strong sign game at the White House today
At #ICDM tomorrow: LLM4Sec - Workshop on the use of Large Language Models for Cybersecurity
Program
LLM4Sec Workshop
llm4sec-workshop.github.io
November 12, 2025 at 11:46 AM
At #ICDM tomorrow: LLM4Sec - Workshop on the use of Large Language Models for Cybersecurity
At #ICDM on Friday morning
JCSIC - Workshop (GaLM) | Indian Institute of Technology (BHU)
Graph-Augmented LLMs (GaLM): Bridging Language and Structured Knowledge ICDM 2025 Workshop • November 14, 2025 About the Workshop
iitbhu.ac.in
November 12, 2025 at 11:43 AM
At #ICDM on Friday morning
So the Casimir effect, in which the pressure caused by ambient waves with long wavelengths push two nearby things together (where these waves can’t exist), is a metaphor for camaraderie. I have seen this effect (albeit with deep exceptions) in Ukraine and I hope it happens in America.
A water wave analog of the Casimir effect
Two rigid plates are vertically suspended by thread such that they are parallel to and opposite each other. The plates are partially submerged in a dish of liqu
pubs.aip.org
November 11, 2025 at 6:58 PM
So the Casimir effect, in which the pressure caused by ambient waves with long wavelengths push two nearby things together (where these waves can’t exist), is a metaphor for camaraderie. I have seen this effect (albeit with deep exceptions) in Ukraine and I hope it happens in America.
This stuff is a ton of fun soft condensed matter physics that someday will turn into a ton of fun optics
Scientists built these tiny diamond crystals using a technique known as DNA origami, in which #DNA molecules fold themselves into elaborate shapes.
Learn more in this 2024 #SciencePerspective on #OrigamiDay: https://scim.ag/4qQei75
Learn more in this 2024 #SciencePerspective on #OrigamiDay: https://scim.ag/4qQei75
Engineering colloidal crystals molecule by molecule
DNA particles are programmed to assemble with precision into complex lattices
scim.ag
November 11, 2025 at 6:41 PM
This stuff is a ton of fun soft condensed matter physics that someday will turn into a ton of fun optics
Most of Masayoshi Son’s career at SoftBank has been frittering away vast sums of other people’s money on overvalued moonshot assets
TOKYO (AP) - Japanese tech giant SoftBank says it has sold all its shares in chip maker Nvidia for $5.8 billion.
$NVDA @apnews.com
$NVDA @apnews.com
November 11, 2025 at 11:55 AM
Most of Masayoshi Son’s career at SoftBank has been frittering away vast sums of other people’s money on overvalued moonshot assets
Claude tryna act real smart and saying a bunch of stuff that doesn’t actually work at all until I finally corner it
November 7, 2025 at 2:26 PM
Claude tryna act real smart and saying a bunch of stuff that doesn’t actually work at all until I finally corner it
Reposted by Steve Huntsman
Finding new ways to break ChatGPT
November 7, 2025 at 12:14 AM
Finding new ways to break ChatGPT
Reposted by Steve Huntsman
A new paper with Bogdan Georgiev, Javier Gomez-Serrano, and Adam Zsolt Wagner: "Mathematical exploration and discovery at scale" arxiv.org/abs/2511.02864. Further discussion is at terrytao.wordpress.com/2025/11/05/m...
Mathematical exploration and discovery at scale
AlphaEvolve is a generic evolutionary coding agent that combines the generative capabilities of LLMs with automated evaluation in an iterative evolutionary framework that proposes, tests, and refines ...
arxiv.org
November 6, 2025 at 3:42 AM
A new paper with Bogdan Georgiev, Javier Gomez-Serrano, and Adam Zsolt Wagner: "Mathematical exploration and discovery at scale" arxiv.org/abs/2511.02864. Further discussion is at terrytao.wordpress.com/2025/11/05/m...
www.technologyreview.com/2025/10/30/1...
Um…you can prove it’s not true, and more generally that AGI via learning is impossible. @irisvanrooij.bsky.social did.
Um…you can prove it’s not true, and more generally that AGI via learning is impossible. @irisvanrooij.bsky.social did.
November 3, 2025 at 6:31 PM
www.technologyreview.com/2025/10/30/1...
Um…you can prove it’s not true, and more generally that AGI via learning is impossible. @irisvanrooij.bsky.social did.
Um…you can prove it’s not true, and more generally that AGI via learning is impossible. @irisvanrooij.bsky.social did.
Reposted by Steve Huntsman
I crossed an interesting threshold yesterday, which I think many other mathematicians have been crossing recently as well. In the middle of trying to prove a result, I identified a statement that looked true and that would, if true, be useful to me. 1/3
October 31, 2025 at 7:25 PM
I crossed an interesting threshold yesterday, which I think many other mathematicians have been crossing recently as well. In the middle of trying to prove a result, I identified a statement that looked true and that would, if true, be useful to me. 1/3
Reposted by Steve Huntsman
Instead of trying to prove it, I asked GPT5 about it, and in about 20 seconds received a proof. The proof relied on a lemma that I had not heard of (the statement was a bit outside my main areas), so although I am confident I'd have got there in the end, 2/3
October 31, 2025 at 7:25 PM
Instead of trying to prove it, I asked GPT5 about it, and in about 20 seconds received a proof. The proof relied on a lemma that I had not heard of (the statement was a bit outside my main areas), so although I am confident I'd have got there in the end, 2/3
Reposted by Steve Huntsman
the time it would have taken me would probably have been of order of magnitude an hour (an estimate that comes with quite wide error bars). So it looks as though we have entered the brief but enjoyable era where our research is greatly sped up by AI but AI still needs us. 3/3
October 31, 2025 at 7:25 PM
the time it would have taken me would probably have been of order of magnitude an hour (an estimate that comes with quite wide error bars). So it looks as though we have entered the brief but enjoyable era where our research is greatly sped up by AI but AI still needs us. 3/3
Gift link for first five clicks
Why funding Ukraine is a giant opportunity for Europe
The bill will be huge. It is also a historic bargain
economist.com
October 30, 2025 at 10:47 PM
Gift link for first five clicks
>>>>ASU beating a top 10 team
October 19, 2025 at 12:47 PM
>>>>ASU beating a top 10 team
October 19, 2025 at 1:38 AM
October 18, 2025 at 11:55 PM
October 18, 2025 at 8:26 PM
Reposted by Steve Huntsman
Competitor enters a major AI competition (RNA folding)
GPU poor so can't train an AI
Builds a "classic" eng pipeline instead. (90s tech)
Wins and beat everyone using DL 💀
Their winning "hybrid" model had an AI in it. Their original one did not and had a higher score
So they won despite the AI 😂
GPU poor so can't train an AI
Builds a "classic" eng pipeline instead. (90s tech)
Wins and beat everyone using DL 💀
Their winning "hybrid" model had an AI in it. Their original one did not and had a higher score
So they won despite the AI 😂
While many teams relied on deep learning, the winning team (jaejohn) surprised everyone with a highly optimized pipeline that revived classic template-based modeling. 👇
www.kaggle.com/competitions...
www.kaggle.com/competitions...
1st Place Solution | Kaggle
Hybrid TBM + DRfold2 Approach
www.kaggle.com
October 17, 2025 at 11:19 PM
Competitor enters a major AI competition (RNA folding)
GPU poor so can't train an AI
Builds a "classic" eng pipeline instead. (90s tech)
Wins and beat everyone using DL 💀
Their winning "hybrid" model had an AI in it. Their original one did not and had a higher score
So they won despite the AI 😂
GPU poor so can't train an AI
Builds a "classic" eng pipeline instead. (90s tech)
Wins and beat everyone using DL 💀
Their winning "hybrid" model had an AI in it. Their original one did not and had a higher score
So they won despite the AI 😂
Zermelo-Frankel is inconsistent
In honor of spooky month, share a 4 word horror story that only someone in your profession would understand
I'll go first: Six page commercial lease.
I'll go first: Six page commercial lease.
October 12, 2025 at 8:21 PM
Zermelo-Frankel is inconsistent
Scientists who might bend a knee take note: this guy is a lab head at NIH. Hang together not separately
Seeing a lot of celebration of this as a win.
But at NIH we have seen this exact play before.
It looks like it is the Vought strategy -and to stop the lawless firings we have to see the bigger picture.
A 🧵:
But at NIH we have seen this exact play before.
It looks like it is the Vought strategy -and to stop the lawless firings we have to see the bigger picture.
A 🧵:
Trump Administration Will Rehire Scores of Experts Fired in Error
www.nytimes.com
October 12, 2025 at 12:23 PM
Scientists who might bend a knee take note: this guy is a lab head at NIH. Hang together not separately
This afternoon. Montréal. If you can’t make it here in time then read the paper in the thread
October 10, 2025 at 11:49 AM
This afternoon. Montréal. If you can’t make it here in time then read the paper in the thread