Computational neuroscientist. Founder, https://xitter.cf/t.co/YMBvt487lT: The text-first dating platform. Looking for AI/ML/DS work.
Joined April 2019
- Tweets4K
- Following937
- Followers949
- Likes12K
Pinned Tweet
For the past few years, I've been trying to create a healthier ecosystem for online dating. The idea is that instead of browsing people's pictures, you are first shown their Tile -- a rectangle where you write 7 things about yourself. Join us at NotAZombie.net! And help spread the word!
Toviah Moldwin retweeted
Dendritic computation for abstract rule-based flexible decision-making: science.org/doi/10.1126/scie…
"apical dendrites preferentially encoded the inferred abstract rule and somata preferentially represented sensory stimuli, whereas choice signals were present in both compartments."
Toviah Moldwin retweeted
goddamn I love biophysics
biorxiv.org/content/10.64898…
My take on interpretability from a neuroscience perspective is that the brain (at least in certain regions) tends to have more interpretable single neurons than naively expected from a neural network model, and this is a strange feature of the brain.
Toviah Moldwin retweeted
A new manhattan project for mech interp named "Make interp" that gets them to stop writing 20 pages of lesswrongese purple prose without ever getting to the point.
I'm not sure I agree that it's evil, there might be a scenario where everyone is better off. But it's harder to imagine that scenario than it is to imagine it being bad for everyone, because right now it seems bad for everyone.
Almost every AI job doom conversation comes down to the pro-AI side not really accepting the AGI premise. Their idea of AGI implicitly assumes it will lack some capability that humans have. Which may be what happens! It is possible that humans will maintain some advantage. But we don’t know, it’s a huge gamble, and the project of the AI companies is to take those advantages away. It’s an evil project and must be opposed, not because it will certainly succeed, but because it’s wrong to try.
Toviah Moldwin retweeted
'How biological synapses self-assemble gradient learning"
from Tommy's (Poggio) team @mcgovernmit
@qianli_liao Ziyn @LiuZiyin10 , @yule_gan, ..
@PNASNews
pnas.org/doi/10.1073/pnas.26…
The future of ~80% software is backend-only with an API for LLMs. Unless you provide a very good UX, many/most CRUD operations are more conveniently done via an LLM chat interface or an ad hoc UI generated by the LLM.
The pivot will be a mess.
Toviah Moldwin retweeted
Huh.
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
A lot now hinges on whether there are boughs in math and science with fruits that we were not smart enough to pick, or even see. What mysteries of the universe will be unlocked by superintelligence?
Toviah Moldwin retweeted
The backdrop of 2026 Optogenetics method #NobelPrize and how editors of journals play a part in the science enterprise.
Pan's paper got rejected by the editors in nature neuro. Boyden/Deisseroth paper was not (even though it was rejected by Nature and Science). Pan's paper took a longer route and was eventually published in Neuron. That led to massive early funds coming to the side of Deisseroth/Boyden pacing their way to establishing big labs, pushing things further and leaving the competition behind.
Hopefully, in the long run, preprint servers will render the editorial subjective decisions pointless.
scientificamerican.com/artic…
Toviah Moldwin retweeted
I had a feeling recently that seems to be corroborated by chats with colleagues, so I decided to share it here.
A lot of systems neuroscientists driven by algorithmic-level questions are going through what looks and feels like a research depression. 🧵
(1/n)
There should be a law that if a company's interview process takes longer than ~2.5 hours, they need pay the interviewees a per-hour consultancy rate for any additional time spent.
Questionably popular opinion: @AmazonKindle should have an LLM built in to ask it questions about what you're reading.