...

OpenAI researchers recently claimed a major math breakthrough on X, but quickly walked it back after criticism from the community, including Deepmind CEO Demis Hassabis, who called out the sloppy communication.

For Richard, a crazy idea was an opportunity to either prove it wrong or prove it right.

Note: If you’ll forgive the shameless self-promotion, applications for my MATS stream are open until Sept 12. I help people write a mech interp paper…

Hobbyist training AI on Victorian texts gets an unexpected history lesson from his own creation.

Instructions in preprints from 14 universities highlight controversy on AI in peer review

Posted on Thursday 3 Jul 2025. 1,012 words, 10 links. By Matt Webb.

Introducing a new, unifying DNA sequence model that advances regulatory variant-effect prediction and promises to shed new light on genome function — now available via API.

Personal Blog

5 years ago, nobody would have guessed that scaling up LLMs would as successful as they are. This belief, in part, was due to the fact that all known statistical learning theory predicted that massively oversized models should overfit, and hence perform worse than smaller models. Yet the undeniable fact is that modern LLMs do possess models of the world that allow them to generalize beyond their training data. Why do larger models generalize better than smaller models? Why does training a model to predict internet text cause it to develop world models? Come deep dive into the inner working of neural network training to understand why scaling LLMs works so damn well. Want to see more videos like this in the future? Support me on Ko-fi https://ko-fi.com/algorithmicsimplicity Papers referenced: Double Descent: https://arxiv.org/abs/1812.11118 The Lottery Ticket Hypothesis: https://arxiv.org/abs/1803.03635 My previous videos on Autoregressive Transformers: Auto-regression (and diffusion): https://youtu.be/zc5NTeJbk-k Transformers: https://youtu.be/kWLed8o5M2Y

Implementation of the Aurora model for Earth system forecasting - microsoft/aurora

A comprehensive suite of tools, built to liberate science by making the creation, evaluation, and dissemination of research more transparent, affordable, and efficient. - robertjakob/rigorous

I used AI in my plasma physics research and it didn’t go the way I expected.

Once errors creep into the AI knowledge base, they can be very hard to get out.

Therapeutics Commons (TDC): Multimodal Foundation for Therapeutic Science - mims-harvard/TDC

We asked some of the world’s foremost experts an impossible question. Amazingly, they answered.

Artificial intelligence foundation for therapeutic science

What exactly is consciousness, and why is it such a hard problem to solve? Neil deGrasse Tyson and co-hosts Chuck Nice and Gary O’Reilly take you deep into the mysteries of consciousness and objective reality, David Chalmers, a philosopher and cognitive scientist. How do we understand the core of subjective experience? We explore the philosophical and neuroscientific angles of consciousness. Why do some experts believe consciousness arises from the sensory cortex, while others point to the prefrontal cortex? And what does this mean for how we understand animals, AI, or even the very nature of our own minds? We question how consciousness evolved to discussions about the levels of consciousness in animals, babies, and humans under anesthesia. As the conversation turns to AI, the group contemplates the future of artificial consciousness. Could large language models, like today's AI, eventually develop consciousness? If not now, will they in the near future? And if we can simulate human consciousness in machines, are we creating philosophical zombies—entities that behave like us but lack awareness? These questions spiral into debates about reality, VR, and whether we're already living in a simulation. Could there be laws of consciousness waiting to be discovered, like the laws of physics? Or is consciousness simply an illusion, as some scientists propose? What are the core features of consciousness that scientists agree on? Buckle up for a fascinating journey that will leave you questioning not just what consciousness is, but whether any of this is even real. Thanks to our Patrons Jay, Gregory Aronoff, Tom B. Night, Barnsley, Glenn, Hibachi Flamethrower, Crescencio Maximilian joseph Martinez, Micheal Gomez, Matthew Deane, James, Joe Chillemi, Thomas van Cleave, Kelsey Plugge, Jeff Jones, William Hamilton, and Kevin Cosg. for supporting us this week. Check out our second channel, @StarTalkPlus Get the NEW StarTalk book, 'To Infinity and Beyond: A Journey of Cosmic Discovery' on Amazon: https://amzn.to/3PL0NFn Support us on Patreon: https://www.patreon.com/startalkradio FOLLOW or SUBSCRIBE to StarTalk: Twitter: http://twitter.com/startalkradio Facebook: https://www.facebook.com/StarTalk Instagram: https://www.instagram.com/startalk About StarTalk: Science meets pop culture on StarTalk! Astrophysicist & Hayden Planetarium director Neil deGrasse Tyson, his comic co-hosts, guest celebrities & scientists discuss astronomy, physics, and everything else about life in the universe. Keep Looking Up! #StarTalk #neildegrassetyson Timestamps: 00:00 - Introduction: David Chalmers 04:00 - What is Consciousness? 06:50 - Where is the Seat of Consciousness? 14:30 - Are Large Language Models Conscious? 22:47 - The Laws of Consciousness 31:40 - Does Consciousness Really Exist? 34:50 - The Reality of Virtual Reality 42:40 - The Future of Reality

ESM3. Enabling scientists to understand, imagine, and create proteins.

List of molecular design using Generative AI and Deep Learning - GitHub - AspirinCode/papers-for-molecular-design-using-DL: List of molecular design using Generative AI and Deep Learning

What happens when big advances in life science converge with A.I.

The 2000-year-old scroll discusses music, food, and how to enjoy life’s pleasures.

For over two decades, many people, including me , have been writing programs attempting to replicate "medicinal chemistry intuition".   This...

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

arXiv dataset and metadata of 1.7M+ scholarly papers across STEM

A great resource for all scientists.

How to predict protein functions using the CAFA2 Competition Methodology — Part I