Pinned Tweet
It's one thing to say what is not working in statistical methods (aka modern AI). I go further and propose an alternative that may address those issues. I am open to collaboration.
Paper: philarchive.org/rec/NAUNOC-2
Posts: alexandernaumenko.substack.c…
Video: youtube.com/watch?v=Uoj-Rz-9…
Alexander Naumenko retweeted
There is a saying in Japanese made from the names of four trees.
Cherry, Japanese apricot (ume), peach, and plum.
All four of them bloom.
But not on the same day.
One in February, one in April.
The cherry doesn't look at the apricot and say 'Im late.'
The peach doesn't compete with the cherry.
Each one blooms in its own time.
And all of them are spring.
What do we do?
Someone our age got married, we're late.
Our friend bought a house, we fell behind.
We keep our calendar by someone else's garden.
The thing is, Your flower isnt late.
Your season just hasnt come yet.
This is called Oubaitori (桜梅桃李)
Japanese · every flower blooms in its own time, don't compare yourself to others.
Is there a saying like this in your language ?
Alexander Naumenko retweeted
Meet the retinal cones. In 1851, Heinrich Müller described how rods and cones differ in shape. Today we know that three types of cones give us colour vision. Packed tightly in the fovea, cones also let us see fine details in bright light, like the tiny print here. #Neuroscience
Alexander Naumenko retweeted
If you wanted a peek inside Morphoceuticals (morphoceuticals.com/) - the startup company seeking to apply our work on bioelectricity to human regeneration and repair, here it is:
lnkd.in/p/gxf5Njzi
This is just a very short teaser. They're doing some amazing stuff - much more coming in the next year. @KaplanLab_Tufts and I are the scientific co-founders (though we're not employees of the company, we work closely together to bring our basic research toward pipelines of human health). The team has made some very cool and fundamental advances over the last couple of years, in reading and understanding the bioelectrics of clinically-relevant model systems. Really excited about it!
Alexander Naumenko retweeted
For decades, a grisly mystery has buzzed around one fly-killing fungi: No matter when the fungus infects a fly, it always dies at sunset. Now scientists have discovered why. sciencenews.org/article/fly-…
"Knowledge" is a vague term. "Difference" is a term guiding implementation.
Compare the quote below with "What else makes a difference?"
Replying to @haider1
As Jean Piaget is famously quoted as having said: "Intelligence is not what you know. It is what you do when you don't know"
5
334
It all boils down to the categorization of LLMs' output. I claim it's a word salad. Language generated by humans relies not on the statistics of the training corpus, but on the scene and properties of relevant phenomena. There's a difference!
Full-blown language sets us apart from other animals, so we naturally assumed that it was the capstone of the intelligence curriculum, with perception and embodiment as prerequisites. But LLMs suggest it's more of a standalone 101 course, and we're just apes who double-majored.
9
225
Replying to @AlexanderNaume2
Offline cellular housekeeping. Evolution cornered itself: heavy neural maintenance doesn’t function when the system is actively firing.
Who knows a good theory for why all species sleep?
If evolution preserves it, there must be a critical reason for it.
Alexander Naumenko retweeted
Meet the rods that let you see under starlight. Treviranus described rod-shaped cells in 1828, and Heinrich Müller described how rods and cones are arranged in 1851. In 1979, Baylor, Lamb, and Yau showed that a single photon can trigger a response in a toad rod. #Neuroscience
Alexander Naumenko retweeted
The tongue of a woodpecker wraps around its skull to protect its brain while pecking
General misconception.
Language does not transfer information, it points to the source, other modules retrieve information there, restoring the "losses."
Alexander Naumenko retweeted
Each octopus arm is so packed with neurons it can make decisions independently from the central brain, and sometimes steal food from its other arms.
Alexander Naumenko retweeted
A groundbreaking @UCSF study from @ChangLabUcsf, published in @ScienceMagazine, identified a subregion of Wernicke's area becomes active before we speak and is necessary for speech production—encoding features of an upcoming word.






























