🧠 computational neuroscience | emergence 🤖 Research Staff Scientist at Neuroblox, Inc. (views here are my own!)
Boston, MA
Joined March 2009
- Tweets31.8K
- Following954
- Followers7.5K
- Likes22.2K
Pinned Tweet
Here's the latest installment in my series on neuroscience and the hyperreal. I look at the potential dangers of "walled garden thinking", and why extensible modeling is important.
Yohan J. John retweeted
This is a beautiful dictionary of the different kinds of Goodhart proxy failures across different domains! Love these kind of interdisciplinary work.
Super excited to announce that I've joined Neuroblox as a Research Staff Scientist! We're building a multi-scale comp neuro platform for mechanistic modeling of cognition, behavior and mood.
neuroblox.ai/
Yohan J. John retweeted
Like @davidbessis and others, I think that Hinton is wrong. To explain why, let me tell you a brief story.
About a decade ago, in 2017, I developed an automated theorem-proving framework that was ultimately integrated into Mathematica (see: youtube.com/watch?v=mMaid2jY…) (1/15)
Here's the next part of my series on neuroscience and hyperreality.
I look at the distinction between forward and reverse modeling.
Yohan J. John retweeted
Very sad to learn of the death on 18th July of Margaret (Maggie) Boden, a titan of cognitive science and AI. I met her many times, and respected her greatly.
theargus.co.uk/memorials/dea…
ALT Maggie Boden and Murray Shanahan on a panel in 2018
Today I learned that the manifold hypothesis might be traceable all the way back to Helmholtz.
From 'Operationalizing Kant Manifolds, Models, and Mathematics in Helmholtz's Theories of Perception'
doi.org/10.7551/mitpress/404…
I'm starting a series of essays that view neuroscience through the lens of Baudrillard's concept of hyperreality.
The first part is out now.
Yohan J. John retweeted
I love that it's not just "turtles all the way down", but turtles all the way up, down, left and right, filling up the entire plane. The reflected turtles get three stripes, as they are where three lines meet.
The proofs are over my head, but I still really love this paper. It also provides a nice metaphor for understanding emergence in a system that is completely "transparent" (since it is an abstract mathematical phenomenon).
1/n
arxiv.org/abs/2303.10798
In which I connect some of the dots from BF Skinner to LLMs! :P
Have you ever wondered how reinforcement learning (RL) is used to train large language models (LLMs)?
Check out the latest Kempner Byte, by @DrYohanJohn, which offers an intro to RL and explains how RL unlocks the potential of LLMs!
kempnerinstitute.harvard.edu…
#RL #ML #AI #LLMs
Recently discovered a band called Gauley Bhai, from Kalimpong and Bangalore. Their new album 'Sunbari' is excellent. Nepali folk-rock with touches of Tuareg desert blues.
youtu.be/Nn_u6F2wYUQ?si=yO-W…
"... those who stayed shifted their attentions away from original research and toward hermeneutics: interpreting the scientific work of metahumans."
Ted Chiang wrote this in the year 2000 in a sci fi essay in Nature (!?). Reminds me of research on interpretability in ML. 😛
I had an interesting conversation with Sam Gershman recently.
Check it out 👇
Check out the new Q&A with Kempner associate faculty member Sam Gershman to get a glimpse of his research on biological intelligence — including some exciting work on non-synaptic learning!
bit.ly/KempnerGershman
@PsychHarvard @gershbrain
#neuroscience #psychology #cogsci
"LMCompress shatters all previous lossless compression records on four media types: text, images, video and audio."
nature.com/articles/s42256-0…
Andor is a great example of a phenomenon we could call the Hegelian suspension of older genres in new ones.
It's easy to discern the WW2 and Cold War spy thriller vibes suspended in (and revivifying?) a dessicated sci fi exoskeleton.
"People outside the field are often surprised and alarmed to learn that we do not understand how our own AI creations work. They are right to be concerned: this lack of understanding is essentially unprecedented in the history of technology."
The Urgency of Interpretability: Why it's crucial that we understand how AI models work darioamodei.com/post/the-urg…
