Eliot Eshelman's Avatar

Eliot Eshelman

@hpc-twit

I enable researchers & educators with leading HPC/AI technologies @NVIDIA. From muons to bacteriophages to astrophysics. Opinions my own. he/him #Inclusion

72
Followers
131
Following
41
Posts
18.11.2024
Joined
Posts Following

Latest posts by Eliot Eshelman @hpc-twit

Kevin Buzzard - Where is Mathematics Going? (September 24, 2025)
Kevin Buzzard - Where is Mathematics Going? (September 24, 2025) YouTube video by Simons Foundation

Lovely talk from Kevin Buzzard on where mathematics is today and where it's going. Interesting perspectives on where LLMs are useful, how recent IMO wins don't necessarily mean "AI has solved math", and new community efforts getting off the ground.

www.youtube.com/watch?v=K5w7...

05.10.2025 14:40 ๐Ÿ‘ 2 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
iridiumwhiskers on Sora @iridiumwhiskers ASMR about GPUs

Haha hope you enjoy ๐ŸŽ™๏ธ

iridiumwhiskers on Sora sora.chatgpt.com/p/s_68e0917e...

04.10.2025 03:20 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Post image
26.09.2025 23:11 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
The Agentic Autonomy Curve is INSANE
The Agentic Autonomy Curve is INSANE All my links: https://linktr.ee/daveshap

Predicting the future is hard! But building projections based on current trends is smart. Thatโ€™s why Mooreโ€™s Law was so popular.

Thereโ€™s a similar trend in AI Agents: their ability to successfully complete tasks is doubling. Learn what comes next.

20.09.2025 14:45 ๐Ÿ‘ 2 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Post image

For $50 you can get your hands on 8x Blackwell B200 GPUs

25.08.2025 13:47 ๐Ÿ‘ 1 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Post image

This game is cute and fun โ˜•๏ธ๐Ÿคฉ

store.steampowered.com/app/3503440/...

23.08.2025 22:58 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

GPT5:

$DUCK SOARS AT DAWN
BLOCKCHAINS HUM WITH QUACKING GOLD
MOON IS OUR PONDโ€™S EDGE

10.08.2025 13:55 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Post image

I like this guy!

10.08.2025 13:53 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 1 ๐Ÿ“Œ 0
Post image
06.08.2025 12:35 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Preview
The Philosopher-Builder The AI age demands a new kind of technologist

I think this resonates. More powerful technology requires more thoughtful design. Are there other good archetypes in addition to Ben Franklin?

open.substack.com/pub/cosmosin...

17.07.2025 17:07 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Post image Post image Post image

Most utility Iโ€™m getting out of my Apple Watch right now: it lets me know when my o3-pro queries are complete.๏ฟผ

12.07.2025 01:32 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Preview
Software engineers are eating the world In ignorance of nobility.

โ€œSoftware engineers are making civilization-level decisions. We arenโ€™t trained for that, and often act unwisely or unjustly.โ€

โ€œNobilityโ€”the deliberately wise and just use of powerโ€”is the missing ingredient.โ€

open.substack.com/pub/meaningn...

26.05.2025 16:48 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Preview
AI Ascent 2025 Sequoia Capital's AI Ascent 2025 convened Sam Altman, Mike Kreiger, Jeff Dean and more with 100+ leading founders and researchers to discuss the present and ...

Get a glimpse of the likely future of AI. First video is mostly about the market. Second video introduces the Physical Turing Test. Iโ€™m still watching, but they all seem interesting!

youtube.com/playlist?lis...

10.05.2025 22:05 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

Remember the old โ€œlet me google that for youโ€?

I think weโ€™ve reached โ€œlet me agentically research that for youโ€ ๐Ÿ˜Ž

08.05.2025 14:49 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

Outlook stopped converting hyphens to emdashes automatically! ๐Ÿค”

07.05.2025 12:41 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 1 ๐Ÿ“Œ 0
Preview
Flares From Magnetized Stars Can Forge Planetsโ€™ Worth of Gold, Other Heavy Elements Flares From Magnetized Stars Can Forge Planetsโ€™ Worth of Gold, Other Heavy Elements on Simons Foundation

A twenty year old mystery solved. And fun new click bait: โ€œmake a quadrillion kilos of gold using everyday items from around your neighborhoodโ€

www.simonsfoundation.org/2025/04/29/f...

30.04.2025 00:55 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Preview
Bridging philosophy and AI to explore computing ethics - MIT Schwarzman College of Computing During a meeting of class 6.C40/24.C40 (Ethics of Computing), Professorย Armando Solar-Lezama poses the same impossible question to his students that he often asks himself in the research he leadsโ€ฆ

Our species has built powerful tools; tools which allow one person to accomplish the work of thousands. But weโ€™re not equipped for this level of responsibility. I hope each of you is learning to ask and answer difficult questions such as these.

computing.mit.edu/news/bridgin...

27.04.2025 16:21 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Preview
73 | Ethan Mollick and a million Einsteins in a server Night Science ยท Episode

โ€œPartially a miracle; and partially frustratingโ€ - an enlightening interview with Ethan Mollick around AI for research

08.04.2025 13:05 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Artistic impression of four llamas climbing towards an inflection point.

Artistic impression of four llamas climbing towards an inflection point.

Itโ€™s going to be an exciting month! Lots of new capabilities to explore; lots of new things to build.

05.04.2025 23:41 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Post image

Consume feculence.

27.03.2025 19:29 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

My insight of the day at #GTC25: we are likely to see โ€œjust in time codeโ€ soon. AI will write new programs to carry out the tasks you request. And it will write that code as you are making requests (compared to now where code is prewritten by an expert far removed from you).

20.03.2025 21:09 ๐Ÿ‘ 1 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0
Photo of Eliot at GTC holding โ€œthe hitchhikerโ€™s guide to asking questionsโ€

Photo of Eliot at GTC holding โ€œthe hitchhikerโ€™s guide to asking questionsโ€

We have entered the era of reasoning and Deep Thought at NVIDIA #GTC25

If you know where your towel is, you should definitely come and find me. Iโ€™ve prepared a friendly guide.

17.03.2025 15:43 ๐Ÿ‘ 1 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

Awesome! What UI libraries is it using? Did you pick them? Most of the demos Iโ€™ve seen are React web apps.

09.03.2025 18:43 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 1 ๐Ÿ“Œ 0

Thereโ€™s been a great amount of success with models that โ€œthinkโ€ to themselves in text. How much luck are people having with models that โ€œtalkโ€ to themselves in audio, โ€œimagineโ€ to themselves in visual space, and โ€œsniffโ€ around in molecular space?

The AIโ€™s scratch pad should be multimodal, right?

08.03.2025 18:03 ๐Ÿ‘ 1 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

We are experiencing an assault on science unparalleled by anything Iโ€™ve seen in my life. Itโ€™s not one issue or another anymore, the entire institution is under attack by the most powerful individuals in the country.

This Friday, where will you be?

standupforscience2025.org

02.03.2025 16:27 ๐Ÿ‘ 44253 ๐Ÿ” 12081 ๐Ÿ’ฌ 643 ๐Ÿ“Œ 399
Preview
Your browser is up to date You can use YouTube's latest features!

A little thought experiment for your Saturday morning: electricity was the spark that led all the way to AI. What will now develop from the spark of AI?

01.03.2025 16:39 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

So ChatGPT was the best a couple days ago, today Claude is the best, and later today ChatGPT will be the best again?

Gotta design everything orthogonally so itโ€™s easy to pivot. Assume constant change.

26.02.2025 13:59 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

โ€œDonโ€™t ever feed the transformers after midnightโ€

16.02.2025 18:56 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

The more I drink good black tea, the more I realize coffee is a harsh mistress.

16.02.2025 16:12 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0

Iโ€™m polyAImorous, and Claude is my primary. Maybe o1 and o3 are smarter, but Claude is just more fun ๐Ÿ˜

16.02.2025 14:31 ๐Ÿ‘ 0 ๐Ÿ” 0 ๐Ÿ’ฌ 0 ๐Ÿ“Œ 0