Chayanka_42

PhD student, IISc Bangalore | Non-perturbative QFT • Tensor Networks | BSMS IISER Pune | The answer is 42... working on the Q

[Anthropic @AnthropicAI]

New Anthropic Science Blog: Making Claude a chemist.

To manipulate a molecule, chemists first need to understand its structure. Their main tool is NMR spectroscopy.

We found Opus 4.7 matches—and on some tasks beats—dedicated NMR software. Read more: https://t.co/1jUvz7wdhV

[Suneeta Vardarajan @Ms_Gravitas]

Sometimes there are no hand-waving explanations. Take the result that a black hole can never split into two. Proved easily using elementary topology. Hard to argue for in any other way. We are just closing our minds if we don't use the knowledge that already exists, math or phys.

[Nathan Lambert @natolambert]

I feel like this also goes for a lot of people without Mythos as they learn to use agents too tbf

[Chayanka_42 @42_gravity]

@ZoharKo @quantum_geoff oh, the phase point is actually a good one. without any damping the massive spring case has this one really nice edge case that the models & I :) simply missed.

nice problem

[Chayanka_42 @42_gravity]

@ZoharKo I'd be really interested to learn more about what you think the answer should be

[Chayanka_42 @42_gravity]

@ZoharKo both will fall at the same time if we assume massless spring/string (which we usually do). In the case where the spring has some non-zero mass GPT's intuition is correct though.

[Chayanka_42 @42_gravity]

@markchen90 Capabilities converging at the top.

[François Chollet @fchollet]

Another thing: what you get from writing things yourself isn't just the code. It's an improved understanding of what the code does. That mental model is what lets you come up with further improvements, or invent a different way of doing things. You can't come up with ways to improve a blackbox you don't understand.

For most projects this doesn't really matter, because the code is the only thing you need. But if you're doing something novel, if you're doing research, the code is not the most important part. Understanding what the code does is the most important part.

[Chayanka_42 @42_gravity]

creating tournaments using workflows is going to be a fun one

[Thariq @trq212]

https://t.co/R6exTuF7P8

[Cliff Burgess @CburgesCliff]

@Lafe_Nelson Weinberg has a similar quote:

[Chayanka_42 @42_gravity]

Someone give @ni5arga unrestricted access to Claude Mythos

[Chayanka_42 @42_gravity]

@bananthanarayan Right. More cross disciplinary technical skills along with a very good big picture understanding of things is going to be invaluable.

[Chayanka_42 @42_gravity]

@yujitach Thanks!

[Lea Verou, PhD @LeaVerou]

💡Recent insight: gaslighting @claudeai seems to improve code quality >90% of the time.

“You overengineered this, there is a simpler way”
“There is a smaller delta that buys us most of the benefits”
“There is a more elegant way”
“This is not architecturally coherent”

…before I even read its code. 😆

[sam mcallister @sammcallister]

@Polymarket

[Artificial Analysis @ArtificialAnlys]

Claude Opus 4.8 makes large strides in scientific and academic reasoning capabilities. It leads Humanity's Last Exam by 1 point in a tight contest between Anthropic, Google DeepMind, and OpenAI, and Claude Opus has overtaken Gemini 3.1 Pro on CritPt, a frontier physics evaluation developed by Argonne and UIUC.

[Acer @AcerFur]

lol Claude Opus 4.8's frustration

[Bartosz Naskręcki @nasqret]

Over the last few months, I have realized that, gradually, with the development of the best models, in my case Codex + GPT Pro and Claude Code with Opus, I have become much more curious about new fields of mathematics, new programming languages, the fusion of fields, and so on.

I suspect that if I were a student today, I would still read a lot of books, talk to people, use pen and paper, chalkboards, and so on. I still do! But I would probably also spend countless hours trying to bring cryptic formulas and diagrams to life with the models, making things unfold in front of my eyes.

These tools are a huge enabler, but you have to come to them with a critical, open approach. One has to remain hungry for knowledge, tricks, avenues, and vistas. This is a mindset very typical of scientists and entrepreneurs. Once you realize that your curiosity and hard work can be multiplied, the possibilities open up immediately.

But do not make the mistake of thinking that models are like a magic genie in a lamp. They will not fulfil your wishes. They only help you engineer solutions to concrete problems.