Prabhas

Cool

[Prabhas @prabhas]

@elonmusk Yes, this should be the new hope AI must bring for humanity, not an amplifier of greed and chaos.

[Prabhas @prabhas]

@mchulet @X Happy to Connect

[Prabhas @prabhas]

Expose AI to unknown problems like

1. Treatment for all diseases be it simple to criticall one's. Creating vaccine takes around 2 years doesnt matter the advancement of AI
2. There is so much unknown underneath oceans ,there is so much to explore, can humans live under water at all ?
3. There so much unknown in space. Find panets ,civilize all the habitable planets.

There is so much unknown to AI, AI excels at what humans documented so far. It hasn't created or invented solutions like famous scientists like newton/Einstein/Ramanujan etc.

People who still know how to use and control AI along with the field excellence (domain knowledge) are probably going to still be in demand.

So learning about problems (domain) is still needed with critical basic fundamentals doesnt matter AI knows them.

[Manisha Mishra @manishamishra24]

Dr. Pratosh at IISc Bengaluru tells his students that rapid advances in AI are commoditizing intellectual labour.

He raises an unsettling question: if companies stop recruiting on campus in the next 5–10 years—even at India’s top universities—what will the purpose of a university education be ?

A must watch video for everyone in tech

[Prabhas @prabhas]

@commonsenseplay What about NFLX?

[Carlos E. Perez @IntuitMachine]

MIT researchers have been digging into the "brains" of 60 different scientific AI models, and have stumbled upon something wild.

It turns out, whether an AI is reading text or looking at 3D atoms, they are all starting to agree on the same hidden truth about our universe.

Here is the pattern you can't unsee. 🧵

1/
First, the premise.
We have AI models for everything now.
• Some read protein sequences (like text).
• Some look at 3D crystal structures (like vision).
• Some predict forces in materials.

They are built differently. They are trained differently. They should think differently.

2/
But a new paper from MIT just asked a massive question:
"Are these models actually learning the same physics?"
The answer is yes. And it’s kind of spooky.

3/
The researchers took nearly 60 models—from LLMs reading SMILES strings to complex 3D potentials—and peered inside their latent spaces (their internal "thoughts").

They found that as models get smarter, their internal representations of matter start to look identical.

4/
Think of it like this:

If you ask a poet and a physicist to describe a sunset, they use different languages. But if they are both experts, they are describing the exact same reality.

The AI models are converging on a "Universal Representation of Matter."

5/
This chart in the paper is the smoking gun.

It shows that an LLM (trained on text) and a 3D Atomistic Model (trained on geometry) align almost perfectly when looking at molecules.

The text model "hallucinated" the 3D structure implicitly. It learned the physics just by reading the chemistry.

6/
But that's not even the most interesting part.
This convergence gives us a new way to spot "fake" intelligence.

The researchers found that high-performing models all cluster together in this "truth" space.

But the weak models? They scatter.

7/
It’s the Anna Karenina principle of AI:

"All happy (smart) models resemble one another; every unhappy (dumb) model is unhappy in its own way."

If a model diverges from the pack on standard data, it hasn't learned a new trick. It’s just lost in a local sub-optimum.

8/
However, there is a catch.

When the researchers threw "out-of-distribution" data at the weak models (stuff they hadn't seen before), the behavior flipped.

Instead of scattering, the weak models collapsed. They all started making the same low-information mistakes.

9/
This reveals a massive problem in Materials Science AI specifically.

The study shows these models are currently "data-governed." They are memorizing their specific training sets rather than learning universal laws.

They aren't "foundational" yet. They are just really good parrots.

10/
So, what does this mean for the future of Science?

Efficiency: We don't need massive, expensive, symmetry-enforcing architectures. We can "distill" the knowledge from big models into simple, fast ones.

Truth: We can use "alignment" to fact-check AI. If a model disagrees with the consensus of other top models, it's likely wrong.

11/
The most profound takeaway?

pattern-matching
And the fact that different AIs are independently deriving the same laws suggests that these models aren't just pattern matching.

They are uncovering reality.

12/
If this research holds up, in 5 years we won't distinguish between "protein models" and "materials models."

We will just have "Matter Models."
One foundation to simulate it all.

13/
This paper is a dense but rewarding read. It fundamentally changes how I think about "generality" in AI.

If you want to dive deeper, grab the PDF here: [Link to 2512.03750v1.pdf]

And SUBSCRIBE to me for more breakdowns of the science that is quietly changing the world.

[Prabhas @prabhas]

Yes I will vouch the same, I worked on same prototype, by spending around a hundred bucks to generate meaningful knowledge graph before hand. The right chunking of text, with embeddings will auto cluster content in vector db. Knowledge graph is later part,tbf redundant.

[Aamir @aaxsh18]

this is a really bad take. for li, amazon etc. the relationships are well defined and known before constructing the graph. but if it comes to 'memory' for agents you don't want to rely on a graph since relations are unknown beforehand, keeping it up to date is a pain if the relations are semantic, ah and if you use an llm to build the graph you will suffer from models producing different relations for the same thing. don't use graphs, just give the models a really good search

[Prabhas @prabhas]

Hey i m thinking to get same mac but mac studio with 128gb ram. I have few questions on your test

1. Which model you are using for coding
2. Are you using any IDE to do coding with local llm that you are running?
3. How efficient is outcome of code , has it completed tasks you desired?

[Prabhas @prabhas]

@beginnersblog1 @geminiapp is this true?

[Bindu Reddy @bindureddy]

The folks who get maximal benefit from AI and vibe coding are the ones who understand technical concepts and know some basic programming

The can tell when the agent is hallucinating or making mistakes and can nudge it in the right direction

They also understand how to design correctly and prompt it to build what they want

[Akhilesh Mishra @livingdevops]

They say
- DevOps is dead, SRE is dead.
- AI agents will be managing/troubleshooting your Kubernetes clusters.
- Infrastructure-as-Code will be fully automated!
- CI/CD pipelines will be built and managed by AI!

I've been hearing these dramatic predictions since ChatGPT launched.

After 3 years of actually using AI( their top models) in my daily work, I can tell you this: the reality is very different from the headlines.

Yes, AI can write Terraform code(not production ready though), kubernetes manifest(incomplete), basic pipelines and scripts people already posted in github, stack overflow.

But try getting it to:
- Debug a complex Kubernetes networking issue
- Handle multi-region failover scenarios
- Design scalable microservices architecture
- Manage security compliance across cloud providers
- Use newly released, cloud services and security implementation.
- Work with cross teams, negotiating, managing conflicts and keep things running.

Even autonomous AI agents fall short:
- They can't maintain context across your entire infrastructure
- Struggle with real-world edge cases
- Can't understand company-specific requirements
- Limited by their training data when facing novel problems

If your job is just copying Terraform templates, copy-pasting code from Stack Overflow you should be concerned.

But if you understand distributed systems, security implications, and complex infrastructure patterns - AI will amplify your capabilities, not replace them.

The winners will be engineers who can:
- Think deeply about systems architecture
- Solve novel infrastructure challenges
- Use AI to automate routine work
- Focus on high-impact engineering decisions

Stop believing the hype. Start focusing on becoming a better engineer who knows how to use AI as another tool in their arsenal.

The future isn't AI replacing DevOps engineers.

It’s the human who understand technology in depth, remember the issues they faced last year and can make better decision when production is down instead of hallucinating.

[Prabhas @prabhas]

@bradkowalk @naval 💯 UI gets even better. Visual learning/grasp is faster than any other type. Apple is proof. Steve jobs still is right in terms of Visual feel.

[Prabhas @prabhas]

Vision(the literal sense of sight) and format still is big driver for any type of(human/animal/ai) cognition. Can we imagine civilization growth or innovation with out vision so far?

[Naval @naval]

UI is pre-AI.

[Prabhas @prabhas]

@NVIDIAGeForce #GeForceSeason

[Prabhas @prabhas]

@NVIDIAGeForce GeForce Season
Would like to have 2nd gpu to run qwen models for embedding creation
With having single 5090 gpu, I hardly can run both embedding model and qwen3 14b model . Tokens per second for running both quantized models is under 10 tkps. 2nd 5090 will push my RAG progress.

[Prabhas @prabhas]

@elonmusk If everything is cheap, everything around us super intelligent . Then what's left humans to do? Exploring?

Earth is done for humans, what's left is space or under the ocean.

Space is enormous and fun.

You are true visionary 🫡

[Deedy @deedydas]

If you feel like giving up, you must read this never-before-shared story of the creator of PyTorch and ex-VP at Meta, Soumith Chintala.

> from hyderabad public school, but bad at math
> goes to a "tier 2" college in India, VIT in Vellore
> rejected from all 12 universities for US masters despite 1420 on the GRE
> fuckit.jpg
> goes to the US anyway on a J-1 visa to CMU with no plan
> applies for masters (again) to 15 universities
> rejected from all except USC and with late admissions, NYU in 2010
> finds this guy called Yann LeCun (before he was famous)
> starts getting into open source
> rejected from all jobs including DeepMind
> only job is Amazon as test engineer
> his PhD mentor helps him get a job at a small startup (MuseAmi)
> rejected from DeepMind
> couldn't get H-1B because of J-1 home return issue; gets waiver through months of approval with USCIS and US State Dept
> very low on confidence
> In 2011/12 builds one of the fastest AI inference engines on phones
> rejected from DeepMind
> emailed Yann again and joins FAIR because of Torch7 open-source work
> scrapes through bootcamp at Facebook, struggling on an HBase task
> L8/L9 engineers at Facebook struggle to get ImageNet working
> figures out numerics / hyperparam issue as an L4
> first big win!
> FAIR goes well, runs 3 person torch7 team and co-creates PyTorch
> because of politics, management wants to shut down PyTorch
> cries-at-bar.jpg, literally
> eventually some people save PyTorch and it launches in 2017
> gets a EB-1 green card!
> the rest is history...

Think about that. He went to a tier 2 college. Was rejected from all Masters programs 2x. Rejected from every single job except Amazon test engineering. Rejected from DeepMind 3x. Nearly had his baby project shut down. Struggled with visa issues. After 12 years of failures (2005-17), he eventually rose to became a VP at Meta one of the most influential people in AI!

Soumith's story is one of resilience and he's living proof that no matter how down in the dumps you are, there's always hope.

[Prabhas @prabhas]

@TheAhmadOsman @grok explain what's this pewdiepie thingie

[Prabhas @prabhas]

@haider1 @grok if that's the case or if that happens , when would that be in timeline and what all the software Engineers do on earth?

[That Marine Guy🇮🇳 @thatmarineguy21]

An Indian who had been living in Japan for more than a yr noticed something strange : his Japanese friends were polite & helpful, yet none of them ever invited him to their home, not even for a cup of tea.

Confused & hurt, he finally asked one Japanese friend why this was so.

After a long silence, the friend replied, "We are taught Indian history… not for inspiration, but as a warning."

The Indian, astonished, asked, "A warning? Indian history is taught as a warning? Please explain why."

The Japanese friend asked, "How many English ruled India?"

The Indian replied, "Maybe… about 10,000?"

The Japanese person nodded seriously & asked, "At that time, weren’t there over 300 million Indians?"

"So who committed the atrocities on your people? Who followed the orders to whip, torture, & shoot them?"

He asked emphatically, "When General Dyer ordered the firing at Jallianwala Bagh, who pulled the trigger? Was it the English soldiers? No, it was Indians."

"Why didn’t anyone point their rifle at General Dyer, not a single person?" He continued, "The slavery you talk about—this was your real slavery. Not of the body, but of the soul."*

The Indian stood motionless, silent, & ashamed.

The Japanese friend continued, "How many Mughals came from Central Asia? Maybe a few thousand? And yet they ruled you for centuries."

"The Mughals did not rule India through their numbers; it was your own people who bowed to them, obeyed them, betrayed their own, and showed loyalty to the Mughals. Either to survive or for silver coins."

"Your own people converted to their religions."

"Your own people betrayed your heroes to the English. Who betrayed Chandrashekhar Azad? Who informed the English about his hiding place in Alfred Park?"

"Bhagat Singh was not easily executed without the permission of those people (Gandhi-Nehrus) who called themselves patriots."

"You Indians do not need foreign enemies. Your own people repeatedly betray you for power, position, and personal gain. That is why we keep distance from Indians."

"When the English came to Hong Kong and Singapore, not a single local joined their army. But in India, you did not just join the enemy’s army—you served them. You worshipped them. You killed your own people to please them."

"Even today, you have not changed. You have learned no lessons from history. Even for a little free electricity, a bottle of alcohol, or a blanket—you sell your vote, your conscience, and your voice without thinking."

"You chant slogans, protest, but when the country needs your sacrifice, where are you? Your first loyalty is still to your home, family, wife, children, and wealth.

The rest—country —can go to hell."

After saying this, the Japanese left, and the Indian stood there, head bowed, frozen in shame.