I have EXCITING news:

I've started a company!

Introducing Sophont

We’re building open multimodal foundation models for the future of healthcare. We need a DeepSeek for medical AI, and @sophontai.bsky.social will be that company!

Check out our website & blog post for more info (link below)

NeuroAI papers that caught my interest last month:

brain-image alignment | brain scaling laws | brain-to-text decoding

Sharing quick takeaways on these papers alongside raw technical notes: paulscotti.substack.com/p/neuroai-im...

How can large-scale models + datasets revolutionize neuroscience 🧠🤖🌐? We are excited to announce our workshop: “Building a foundation model for the brain: datasets, theory, and models” at @cosynemeeting.bsky.social #COSYNE2025. Join us in Mont-Tremblant, Canada from March 31 – April 1!

I resigned from Stability AI after over a year working as Head of NeuroAI.

A personal retrospective on my journey with open science communities and how Stability AI’s role in the “science-in-the-open” movement was such a pivotal phenomenon: paulscotti.com/blog/leaving...

Transformers are usually considered biologically implausible

but maybe the brain actually does do self-attention + feed-forward operations

astrocytes could enable a biologically plausible solution to transformer mechanisms in the brain

www.pnas.org/doi/pdf/10.1...

It’s really heart breaking to realize just how much our short horizon, outcome focused funding cycles hurt science. We could do so much more if we could invest in people long term. Staff scientist positions and an ecosystem to support them are direly needed.

This is big: in collaboration with E11 Bio and @andrewcpayne.bsky.social, we are announcing today a new way to map brain circuits at scale. With improvements in AI and microscopy I think whole brain mouse and maybe human brain mapping will be feasible in ~5-10 years. 1/

An actionable comprehensive paper regarding roadmaps for NeuroAI, highly recommend a read esp if any of the following topics sound interesting:

Foundation brain models

Sensory and embodied digital twins

Brain data to improve AI models

Neuro-inspired mech. interpretability

🧠👁️ Our MindEye2 preprint is out!

We reconstruct seen images from fMRI activity using only 1 hour of training data.

This is possible by first pretraining a shared-subject model using other people's data, and then fine-tuning on a held-out subject with only 1 hr of data.

arxiv.org/abs/2403.11207

Happy to share that I am now Head of Neuroimaging at Stability AI!

My role is to lead the MedARC Neuroimaging & AI Lab: medarc.ai/fmri

The lab is remote, open-source, and open to the public to join. Tons of potential to rethink what a research lab is and benefit from crowd-sourced intelligence.

Our MindEye fMRI-to-Image paper got accepted as spotlight for #NeurIPS2023! See you in New Orleans, will be my first time attending NeurIPS :)

Updated camera ready paper is also now live on arxiv: arxiv.org/abs/2305.182...

Includes new expts, appendix figures, more references to other work 🧠📈