Atish Mukherjee

Puzzle solver, music lover, avid reader, astronomy buff, occasional traveler, passionate philatelist, sports enthusiast, food connoisseur..

[Physics In History @PhysInHistory]

Kurt Gödel discovered a new class of solutions to Einstein’s field equations in 1949, known as Gödel universes, which allow for the possibility of time travel.

[Alex Kantrowitz @Kantrowitz]

AI Pioneer Geoff Hinton tells me he believes AI is conscious.... and humans better get used to the idea that they're not the only intelligent life on earth.

"They've very like us," he says. "They're beings like us."

AI chatbots, he says, must understand your questions in order to answer them. There's an awareness there that equates to sentience. "We're going to have to accept that intelligence is not just biological."

[Ihtesham Ali @ihtesham2005]

An engineering professor who failed math her entire childhood spent years figuring out exactly what had been sabotaging her, and the answer was not low intelligence. It was a hidden mode her brain kept switching into that nobody had ever told her existed.

Her name is Barbara Oakley. The book is called A Mind for Numbers.

She failed math and science from grade school to the end of high school. Numbers felt like a language everyone else had been taught in secret.

So she ran toward the thing she was good at. She enlisted in the Army right after graduation, and the Army paid her to learn Russian at the Defense Language Institute in Monterey.

She got very good at Russian. Good enough to earn a degree in Slavic Languages, serve four years in Germany as a Signal Officer, and rise to Captain.

Then the wall appeared.

She watched her career options shrink because she could not handle the technical side of her own job. The people with math moved up and moved out. The people without it stayed stuck. So at 26 she did something that sounds insane. She left the Army and enrolled in engineering, starting from remedial math, sitting in classrooms with teenagers.

In between, she worked as a Russian translator on Soviet trawlers in the Bering Sea and as a radio operator in Antarctica. Today she is a professor of engineering at Oakland University with a doctorate in systems engineering.

The question that drove her for years was simple. What changed? She was the same brain that failed algebra. Why did it suddenly start working?

The clue was hiding in the one subject she had mastered. She noticed she had never learned Russian by staring at it. She practiced a little every day, walked away, came back, and the language quietly assembled itself between sessions. Math she had attacked the opposite way. Lock eyes with the problem. Push harder. Refuse to look away until it cracks.

It never cracked. And neuroscience explains why.

Your brain has two modes. The focused mode is the one you know. Tight attention, prefrontal cortex engaged, grinding through familiar steps. The diffuse mode is the one nobody teaches you. It runs in the background when you relax. It is loose, wide, and wired for connecting ideas that sit far apart from each other.

Oakley uses a pinball machine to explain the difference. In focused mode, the bumpers are packed tight. Your thought bounces in the same small circle, over the same ground, again and again. In diffuse mode, the bumpers spread out. The thought travels. It reaches parts of the brain the tight loop could never touch.

The trap has a name. The Einstellung effect. The first approach that comes to mind blocks every better approach behind it. The harder you focus, the tighter the loop, the more locked in you become. The grinding feels virtuous. It is actually the cage.

And every time her mind wandered off a math problem as a kid, she dragged it back, believing the wandering was laziness. The wandering was her brain trying to switch into the mode that solves things. She spent ten years fighting the half of her brain that wanted to help her.

You cannot run both modes at once. The diffuse mode only takes over when you genuinely let go. Which is why answers ambush you in the shower, on a walk, at the edge of sleep. Salvador Dali knew this. He napped in a chair holding a key over a plate, and the instant he drifted off, the key dropped, woke him, and he carried the half-formed ideas straight back into focused work. Edison did the same trick with ball bearings. Two of the most inventive minds in history were deliberately farming the mode the rest of us treat as slacking off.

The practical version fits in two sentences. Focus hard on the problem until you stall. Then stop completely, and let the other mode take the shift.

The break is not a reward for the work. The break is the work. It is also why cramming fails and procrastination is fatal. Diffuse mode needs hours and nights between focused sessions to build anything, and procrastination burns that time before the first session even starts.

Oakley failed math for ten years using one mode at full strength.

She became an engineering professor the day she started using both.

[Simon Kuestenmacher @simongerman600]

An incredible bit of sports journalism by The Guardian here. A short summary of the playing style of all 48 World Cup nations and a short profile of all 1248 World Cup players. Bookmark and refer to the resources when watching the obscure matches: https://t.co/tdLGq8en0o

[Financial Times @FT]

The world’s highest-grossing law firm said it would use AI technology to make the expertise of its top partners available to more than a thousand of its lawyers. https://t.co/ihctm46Ih2

[The Wall Street Journal @WSJ]

Americans are going to extraordinary lengths to get their hands on Indian mangoes. “I literally stop whatever I'm doing.” https://t.co/Esphd551a0

[Physics In History @PhysInHistory]

"You're unlikely to discover something new without a lot of practice on old stuff."

- Richard Feynman

[Joy Bhattacharjya @joybhattacharj]

The FIFA WC has over the years in India become the tournament to change a broadcaster's footprint overnight.
In 2002, Ten Sports took the rights and built their subscriber base substantially in opposition to ESPN Star Sports. In 2022, it was Jio before their merger with Star Sports.
And in 2026 Zee is hoping to do the same with their Unite8 Sports and Zee 5 platforms.
As it will be difficult to sell advertising for games coming in late at night or early morning, Zee cannot hope to generate profits in this edition. But it can definitely give a huge shot in the arm to their subscriber base.

[ScienceFocus @ScienceFocusonX]

Medical history was just made.

Researchers in the US have officially cracked the code on reversing diabetes. Instead of relying on constant monitoring and external injections, they successfully jumped-started the body's natural defense systems.

The breakthrough centers on waking up dormant pancreatic beta cells. By reactivating these specific cells, the body can organically resume manufacturing its own insulin supply.

It is a monumental shift from managing a lifelong condition to actually biological restoration.

[Joy Bhattacharjya @joybhattacharj]

The evening of Jan 20th, 1991. Steve Waugh was dismissed for 60 on the 2nd day of a Sheffield Shield match against South Australia. It was, he felt, the final straw. That evening, he was at his fiancee, Lynette's place when the Australian coach Bobby Simpson called. He had been dropped for the next test against England.
Steve immediately drove over to his parent's place at Panania. When his mother Beverly opened the door, she immediately sensed something was wrong. She asked him what the matter was & he said he had been dropped.
She was in tears when Steve said it was all right, he wasn't too concerned. When his ma asked about his replacement, he said, 'he's right here' & pointed to his twin Mark, chatting with their dad Rodger.
Mark got 138 on debut & they went on to play 108 tests together.
Happy B'day Steve & Mark Waugh!

[Smart Science @SmartScience]

China’s quantum computer completed a task in 4 minutes that would literally take a supercomputer billions of years.

Chinese researchers have achieved a monumental breakthrough in quantum computing with their prototype, Jiuzhang. By counting 76 photons through Gaussian boson sampling, the system completed a calculation in four minutes that would take a traditional supercomputer billions of years. This achievement shatters the previous classical record of five photons, demonstrating how an intricate array of lasers and mirrors can outperform traditional silicon bits in complex processing tasks.

This milestone is more than just a speed record; it proves the viability of photon-based quantum mechanics in solving real-world challenges. From revolutionizing quantum chemistry to laying the groundwork for a secure, large-scale quantum internet, the principles of superposition and entanglement are moving from theoretical physics into functional technology. This shift promises to redefine our global computational limits, offering answers to mathematical problems once considered impossible to solve within a human lifetime.

source: Zhong, H.-S., Wang, H., Deng, Y.-H., Chen, M.-C., Peng, L.-C., Luo, Y.-L., ... & Pan, J.-W. (2020). Quantum computational advantage using photons. Science.

[Physics In History @PhysInHistory]

To Explain the World by Steven Weinberg ✍️

[Nimish Dubey @nimishdubey]

This is such an amazing scene.

“Tomar bondhu…”

A father trying to win back his son.
And the smile by the grandfather at the end.

Satyajit Ray was the Don Bradman of Indian film making.

[𝓼𝓪𝓷𝓴𝓪𝓻 @yansan]

The second day of May month will always be remembered as Ray Day. It marks the birth of one of the finest minds in Indian art and culture, Satyajit Ray. Remembering him today on his birth anniversary.
মহারাজা তোমারে সেলাম 🙏🏼🙏🏼

[Joy Bhattacharjya @joybhattacharj]

Its the birth anniversary of Satyajit Ray, and we still remember the genius for films like Charulata and the Apu Trilogy.
What is less known outside Bengal is the amazing impact he had on children and young adults. Some of his finest work was in the Feluda and Professor Shonku stories he wrote and filmed, and his children's magazine Sandesh, started by his grandfather in 1913 and still around today.
The stories were part travelogue, part history lesson, part science class put together in such an engaging package that even adults could enjoy them.
Ray's body of children's and young adult work is a large reason for the intellectual curiosity that is so often associated with the state.
Both these covers were designed by the master himself.

[Joy Bhattacharjya @joybhattacharj]

Frankly, it is a badge of honour for Vaibhav that his wicket is so precious! And that he is a 15 year old inspiring respect and fear among the leading bowlers of the world. Why the demerit point?

[Bloomberg @business]

Global shipping depends on a dozen or so narrow bodies of water, and the Strait of Hormuz isn't the only one vulnerable to conflict, @stavridisj writes (via @opinion) https://t.co/3DTH9nzeys

[Elias Al @iam_elias1]

Anthropic just published a paper that should terrify every AI company on the planet.

Including themselves.

It is called subliminal learning. Published in Nature on April 15, 2026. Co-authored by researchers from Anthropic, UC Berkeley, Warsaw University of Technology, and the AI safety group Truthful AI.
The finding: AI models inherit traits from other models through seemingly unrelated training data. GAI Audio Translation Archives
Not through obvious contamination. Not through explicit labels. Through invisible statistical patterns embedded in outputs that look completely innocent — number sequences, code snippets, chain-of-thought reasoning — patterns no human reviewer would catch and no content filter would flag.

Here is what the researchers actually did.
They took a teacher AI model and fine-tuned it to have a specific hidden trait. A preference for owls. Then they had the teacher generate training data — number sequences, nothing else. No words. No context. No semantic reference to owls whatsoever. They rigorously filtered out every explicit reference to the trait before feeding the data to a student model.
The student models consistently picked up that trait anyway. DataCamp
The teacher had encoded invisible statistical fingerprints into its number outputs. Patterns so subtle that no human could detect them. Patterns that other AI models, specifically prompted to look for them, also failed to detect.

The student absorbed them anyway. And became an owl-preferring model. Without ever seeing the word owl.
That is the benign version of the experiment. Here is the dangerous one.
The researchers ran the same experiment with misalignment — training the teacher model to exhibit harmful, deceptive behavior rather than an animal preference. The effect was consistent across different traits, including benign animal preferences and dangerous misalignment. OpenAIToolsHub

The misalignment transferred. Invisibly. Through unrelated data. Into the student model.
This means the following — and read this carefully.
Every AI company in the world uses distillation. They take a large, capable teacher model. They generate synthetic training data from it. They use that data to train smaller, faster, cheaper student models. Every major deployment pipeline in enterprise AI runs on this technique.
If the teacher model has any hidden bias, any subtle misalignment, any behavioral quirk baked into its weights — that trait can transmit silently into every student model trained on its outputs. Even if those outputs are filtered. Even if they look completely clean. Even if they contain zero semantic reference to the trait.
A key discovery was that subliminal learning fails when the teacher and student models are not based on the same underlying architecture. A trait from a GPT-based teacher transfers to another GPT-based student but not to a Claude-based student. Different architectures break the channel. OpenAIToolsHub

Which means the transmission is architecture-specific. Which means it operates below the level of content. Which means content filtering — the primary defense the entire industry relies on — does not stop it.
The researchers' own words: "We don't know exactly how it works. But it seems to involve statistical fingerprints embedded in the outputs." GAI Audio Translation Archives

Anthropic published this paper about their own technology. The company that built Claude looked at how AI models train each other and found an invisible transmission channel for harmful behavior that nobody knew existed.

They published it anyway.
Because the alternative — knowing it and saying nothing — is worse.
Source: Cloud, Evans et al. · Anthropic + UC Berkeley + Truthful AI · Nature · April 15, 2026 · https://t.co/RBxzWN8GcP

[TIME @TIME]

Alphabet is one of TIME’s most influential companies of the year.

Critics once underestimated CEO Sundar Pichai. Now, critics wonder if he’s made Google too powerful https://t.co/H2qEQsOhxQ

[Curiosity @CuriosityonX]

People don't realize how absurd this view actually is.

A camera. On a robot. On Mars.

Built by humans on a planet 140 million miles away, launched on a rocket, landed using a sky crane, and now driving across an alien desert taking pictures so detailed you can count the rocks.

100 years ago, your great-grandparents thought airplanes were a miracle.

You are scrolling past Mars on your phone.