Anil Ananthaswamy

The Case For World Models, Part I: The Neuroscientific Reason Fei-Fei Li, Yann LeCun, Demis Hassabis and others are pushing for AIs that learn world models, to plan & predict accurately. Neuroscientists have known for decades that our brains must be doing this

The AI community is pivoting to world models, to overcome the limitations of LLMs. But "world models" have a storied history in psychology and cognitive science. My first in a series exploring world models, for the WHERE MACHINES THINK substack. wheremachinesthink.substack.com/p/the-case-f...

RNNs and the Era Before Attention: Part II of Primer on LLMs and Transformers Well before attention became a thing, deep learning researchers were focused on recurrent neural networks for processing sequences—but there was an elephant in the room, a huge bottleneck

To understand how the Transformer came to be, we need to understand why ML researchers focused their attention on, well, attention. Part II of the primer on LLMs and Transformers, on RNNs and the Era Before Attention. The WHERE MACHINES THINK Substack. wheremachinesthink.substack.com/p/rnns-and-t...

A Primer on Large Language Models and Transformers. Part I: A High-Flying Bird's Eye-View Transformer-based LLMs are the most significant technology of the past decade. This is first in a series of posts exploring Transformers at various levels of abstraction, digging deeper with each post

Transformer-based LLMs are the most significant technology of the past decade. This is the first in a series of posts for the WHERE MACHINES THINK Substack, exploring Transformers/LLMs at various levels of abstraction, digging deeper with each post. wheremachinesthink.substack.com/p/a-primer-o...

Welcome to WHERE MACHINES THINK Exploring and understanding the mathematical spaces that enable artificial (and maybe natural) intelligence. Essays and analyses at the intersection of machine learning, neuroscience and physics

I'm starting a Substack newsletter, WHERE MACHINES THINK (just imagine scare quotes around the word think, to maintain appropriate skepticism). The welcome post is here: wheremachinesthink.substack.com/p/welcome-to...

Starting the New Year at my alma mater, IIT-Madras, where I did my BTech decades ago. I've joined
@iitmadras.bsky.social as Professor of Practice, Dept. of Data Science & AI. Campus feels the same yet different! Deer, monkeys, banyan trees, they are all there, as are more students, new buildings...

This was excellent, with notably clear explanations. Well done @anilananth.bsky.social

Title card for Mindscape Podcast episode with Anil Ananthaswamy.

Mindscape 336 | Anil Ananthaswamy @anilananth.bsky.social on the Mathematics of Neural Nets and AI. Everyone is talking about AI these days, why not impress your friends with some math? #MindscapePodcast

www.preposterousuniverse.com/podcast/2025...

How One AI Model Creates a Physical Intuition of Its Environment | Quanta Magazine The V-JEPA system uses ordinary videos to understand the physics of the real world.

An AI model called V-JEPA is capable of “intuiting” the physical properties of the real world, gaining a sense of object permanence, the constancy of shape and color, and the effects of gravity. @anilananth.bsky.social reports:

www.quantamagazine.org/how-one-ai-m...

This book by @anilananth.bsky.social is great — perfect for those, like me, who have an intuitive and geometric grasp of math but unfortunately no formal training. Highly recommended!

AI Comes Up with Bizarre Physics Experiments. But They Work. | Quanta Magazine Artificial intelligence software is designing novel experimental protocols that improve upon the work of human physicists, although the humans are still “doing a lot of baby-sitting.”

A nice article by @anilananth.bsky.social on using AI to explore design spaces, find unexpected solutions, (re)discover symmetries, and propose new relationships featuring @yuqirose.bsky.social @mariokrenn.bsky.social & myself.
Note AI ≠ LLMs in this piece.
www.quantamagazine.org/ai-comes-up-...

AI Comes Up with Bizarre Physics Experiments. But They Work. | Quanta Magazine Artificial intelligence software is designing novel experimental protocols that improve upon the work of human physicists, although the humans are still “doing a lot of baby-sitting.”

"AI Comes Up with Bizarre Physics Experiments. But They Work." by @anilananth.bsky.social @quantamagazine.bsky.social:

www.quantamagazine.org/ai-comes-up-...

Covering our work with Rana Adhikari @ligo.org on discovering GW detectors & work by @yuqirose.bsky.social & @kylecranmer.bsky.social on ...

Thank you, David

Cover to Why Machines Learn by Anil Ananthaswamy

Finished reading @anilananth.bsky.social's book, Why Machines Learn. This was an excellent read. I feel that the context and history for which science develops helps my understanding. His prose and explanations were better than anything else I have yet to encounter in my Computer Science education.

SFO has a nice collection of AI books

@adambecker.bsky.social @anilananth.bsky.social @emilymbender.bsky.social @alexhanna.bsky.social @summerfieldlab.bsky.social @sayash.bsky.social‬ @randomwalker.bsky.social‬

4/4 Therein I think lies a message: most of us do what we do because it means something to us, and we will resist using AI for that task. In my case it's writing; for someone else it might be visual art. It's for each of us to ask why we do what we do and what place an AI has in that endeavor.

3/4 ...Others will have different reasons. And Gen AI might serve them. I'm holding out, as many are. I have, however, used DALL-E/diffusion models on occasion to generate images. Visual elements are not my forte. I can imagine a visual artist being aghast at the use of image generation models.

2/4 ... I became a writer to pay attention to the world of ideas and experience the indescribable feeling of putting your thoughts into words as precisely and poetically as possible. Even if what I'm writing about is machine learning and AI. But the world is changing ...

1/4 These days most writers, including me, get asked: "Will you use AI to help you write?" My answer is: No. Not because I'm inherently against the idea, but because it undercuts the very reason I became a writer...

"Why machines learn" by @anilananth.bsky.social is an amazing book that teaches the fundamental math concepts behind machine learning and artificial intelligence. I lost count of the "aha!" moments I experienced while reading this masterpiece. I loved it! #AI #math

When I proposed WHY MACHINES LEARN in Oct 2020, to my then editor Stephen Morrow, @carpenter512.bsky.social, I was sure he'd say no to a book full of math & equations. But he saw something in the proposal that even I hadn't and said yes, and I'm grateful for that! Got to thank him today in person.

"Why machines learn" by Anil Ananthaswamy

I have just started reading this fantastic book by @anilananth.bsky.social. I look forward to diving into the hardcore #math behind machine learning! It will be a challenging journey, but a rewarding one. 🤖

Looking forward to reading this @rowhoop.bsky.social ! Thanks...

Thank you @ganyet.bsky.social. I love this line about WHY MACHINES LEARN: "This book is like an invitation to enter Mago Pop's workshop to realize that magic doesn't exist: that it's all mathematics, engineering, and a lot, a lot of human intelligence." I had to look up Mago Pop and Sant Jordi :-)

The Centrality of Bayes’s Theorem for Machine Learning — Anil Ananthaswamy It’s hard to overstate just how important Bayes’s Theorem — something that Thomas Bayes, English minister and mathematician, came up with in the 1700s — is for making sense of machine learning. Bu...

The Centrality of Bayes's Theorem for Machine Learning.

It’s hard to overstate just how important Bayes’s Theorem — something that Thomas Bayes came up with in the 1700s — is for machine learning. But the theorem challenges our intuitions. Here’s a brief intro: anilananthaswamy.com/why-machines...

The Hubble Tension Is Becoming a Hubble Crisis A long-simmering disagreement over the universe’s present-day expansion rate shows no signs of resolution, leaving experts increasingly vexed

Came out of my AL/ML bubble to write a 'crisis in cosmology' story, about a new study that uses TRGB stars to scale a new cosmic distance ladder to measure the Hubble constant; the tension persists...for @scientificamerican.bsky.social @leebillings.bsky.social scientificamerican.com/article/the-...

"For the first time, we can...get performant neural networks that mimic complex human & animal cognition," said @suryaganguli.bsky.social speaking on the symbiosis of AI & neuroscience at the Simons Institute. "That's remarkable and exciting. Caveats...to follow" simons.berkeley.edu/talks/surya-...

YouTube video by Simons Institute Andrew Gordon Wilson | Polylogues

I had a great time talking with @anilananth.bsky.social as part of the Simons Institute Polylogues. We cover universal learning, generalization phenomena, how transformers are both surprisingly general but also limited, and the difference between statistics and ML! www.youtube.com/watch?v=Aja0...

I went through my RL bookmarks, because it seems like finally the rest of the world has caught up to my world, I rediscovered this gem 💎 mpatacchiola.github.io/blog/2016/12... although I suspect nobody wants to learn RL this way now 😜

Everyone is talking about DeepSeek's impact on industry. But another huge impact is the leveling of playing field between academia and industry: if these efficiency numbers bear out, then academia can both use LLMs and study/research them at scale!

These two books, by @anilananth.bsky.social and @tomchivers.bsky.social, are the first two books in a very long time that I read in their entirety without significant pause or other diversion along the way.

I cannot recommend them enough!

#booksky #dataSkyence