Manav Ponnekanti

Well yeah that’s the situation. Those with the disease are gonna be the ones to search for a cure

YouTube video essays and substacks. Lots of my friends talk about it

It’s already here tbh

The Geometry of Concepts: Sparse Autoencoder Feature Structure Sparse autoencoders have recently produced dictionaries of high-dimensional vectors corresponding to the universe of concepts represented by large language models. We find that this concept universe h...

Interesting. LLMs exhibit modular structures, such as math/code lobes, resembling functional areas in the neocortex.
Just as cortical arealization occurs during development, could insights from developmental neuroscience refine AI architectures? 🧠🤖
arxiv.org/abs/2410.19750

Dopamine and Norepinephrine Differentially Mediate the Exploration–Exploitation Tradeoff Dopamine (DA) and norepinephrine (NE) have been repeatedly implicated in neuropsychiatric vulnerability, in part via their roles in mediating the decision-making processes. Although two neuromodulator...

I think one of the most important challenges in #NeuroAI in the coming years will be mapping the different neuro-modulatory systems onto different internal reward functions!

Papers like this one are the kind of ground work we need for that:

www.jneurosci.org/content/44/4...

🧠📈 🧪