NeuroAI is an interdisciplinary field at the intersection of neuroscience and AI. The brain is energy efficient and robust despite being noisy; by taking our inspiration from the brain, we hope to make AI more robust and efficient. In the other direction, AI tools are helping us peer deeper into the most complex structure known to man, the brain.
There are a number of different focal points for the NeuroAI community. The MAIN conference (Montreal AI and Neuroscience) is a great central location to get a read on the latest in the field. Mila, where I work, has a strong NeuroAI community and hosts a NeuroAI reading group. Paul Middlebrooks hosts the Brain-Inspired podcast. On xcorr, I write about NeuroAI and many other topics.
But if there’s one central hub for NeuroAI discussion, it’s Twitter. That’s where the best discussions are and papers are shared. I’ve met great people in the community on Twitter. But science Twitter is dying. Where is NeuroAI going to convene?
Part of the answer, I hope, is here. I decided to start this Substack as a place to share papers and community news. I’m inspired by AK’s paper shareouts, Davis Blalock’s substack, and the old Nuit Blanche blog which in its heyday was a focus of compressed sensing research. The core format I’m expecting is a compilation of 3-5 papers, with links, key figures, and some comments. I’m hoping for a bi-weekly to monthly cadence.
Why here? Why not on xcorr? A few reasons: first, I think of this as a different format. On xcorr, I wrote long essays. They’ve been getting longer and more polished over time. The latest ones have taken on the order of 40 hours to write. I don’t want to mix long polished essays and ephemera under the same medium. I also want to give myself the space to experiment in a different medium with different constraints. Shorter, rawer, more frequent, off-the-cuff, experimental. I think the substack format is ideal for what I have in mind, because I personally really like to wake up to an email with paper highlights.
I also want to untie this effort from me personally. xcorr is me. Nobody else writes on xcorr. But naix? If would be very cool if there were other contributors in the future.
Here’s hoping this substack will prove a meeting ground for NeuroAI.