Corey
LightNews LightNews
banner
kh0rish.bsky.social
Corey
@kh0rish.bsky.social
560 followers 350 following 5K posts
Biologist, just fed up at this point because *gestures everywhere*
Posts Replies Media Videos
kh0rish.bsky.social
Ive been seeing animorphs and goosebumps get some resurgent love, but whole else read THIS shit right here?
Book cover for Strange Matter, tagline “There is something strange in Fairfield” Purple skies, red moon, and various ethereal figures on the bottom from the various books in the series. Ghosts, werewolves, UFOs and a scary truck.
November 9, 2025 at 7:42 AM
kh0rish.bsky.social
Why are people like this? Autocomplete is literally how the code and the math shakes out. You are making up extra shit to say otherwise
oh man i am pretty tired of the "it's just autocomplete" armchair warriors... it is an accurate metaphor in only the most pedantic of ways, and has zero explanatory power whatsoever as far as intuition building goes. and i don't even understand the impulse. "reality is easy, it's just quantum autocomplete!”
November 8, 2025 at 3:17 AM
kh0rish.bsky.social
Me right now
November 6, 2025 at 9:16 PM
kh0rish.bsky.social
The fuck are SOME of yall doin? Goddamn
Once again, black people lockin in and understanding the assignment. Vast majority (92%) voting against republicans. Other non-asian PoC at least were >60%, but asian folks picked up at 79%. White folks at a miserable 47%
November 5, 2025 at 6:31 AM
kh0rish.bsky.social
I love this one too 😂
3 people in conversation: Person in the middle - "Our field has been struggling with this problem for years." Person on the right holding up a laptop - "Struggle no more! I'm here to solve it with algorithms!" That person is now sat at a desk with the other 2 behind watching him work on the laptop Six months later: Algorithm guy - "Wow this problem is really hard." Original speaker "You don't say"
October 26, 2025 at 6:28 PM
kh0rish.bsky.social
We have never seen that before we dont know what it is its just a skull and crossbones
Christolph Waltz as nazi bastard Col Hans Lands seen here pointing a gun at somewhere off camera, in nazi garb wearing a nazi hat with that nazi totenkopf skull design
October 26, 2025 at 12:55 AM
kh0rish.bsky.social
This goes unbelievably hard what are they feeding these editors man
October 21, 2025 at 8:20 PM
kh0rish.bsky.social
Lol. Lmao even. I forgot about this.
Sam Altman saying on twitter, Feb 3, 2023: eliezer has IMO done more to accelerate AGI than anyone else. certainly he got many of us interested in AGI, helped deepmind get funded at a time when AGI was extremely outside the overton window, was critical in the decision to start openai, etc. it is possible at some point he will deserve the nobel peace prize for this--l continue to think short timelines and slow takeoff is likely the safest quadrant of the short/long timelines and slow/fast takeoff matrix
October 19, 2025 at 10:48 PM
kh0rish.bsky.social
Classic
Iconic Great Moments in Leftism comic about leftist “sectarian infighting” where all the various flavors are holding a sign that says “Unite the Left! United Front” but they are all sneering at each other with thoughts full of contempt
October 19, 2025 at 3:03 PM
kh0rish.bsky.social
Squished himself up
My fluffy black cat Bombur folded in on himself and nestled into my arms Pink Popple stuffed animal toy, which could into itself like a pouch
October 17, 2025 at 9:14 PM
kh0rish.bsky.social
These two assessments go together. If you actually had what you imagine/assume a brain simulation is, you wouldnt need training data, it could just get to analyzing things immediately. It could CREATE models, not be one.
If you have to turn everything into data/math problems before the computer can even operate on it, how are you gonna make an autonomous AI “agent.” You havent come up with the algorithm that turns new encounters and aspects of reality into useable data/math problems. Cells dont have to do that btw Like…guys. Its not a mathematical model of the brain. Its a mathematical model of the dataset. The distribution of word proximity in it, to be precise.
October 11, 2025 at 3:16 PM
kh0rish.bsky.social
Careful, those studies arent definitive and are even in contention when revisiting some patients. The brain yearns for integration
pmc.ncbi.nlm.nih.gov/articles/PMC...
The authors entertain several interpretations of their data, but ultimately, they take the stance that that visual perception remains divided in split-brain patients, but that in reporting what was perceived, consciousness is undivided. They refer to this as "'split phenomenality' combined with 'unity of consciousness" (Pinto et al., 2017a). This interpretation lies in direct contrast with both previous theories of processing in split-brain patients and dominant theories of consciousness. remaining today. It is also worth noting that both patients were tested at least a decade after surgery. In their first experiment, Pinto and colleagues (2017a). examined if the patients could detect a stimulus and indicate its location when presented in only one visual half field. They asked patients to respond with their left hands, right hands, and verbally. Researchers observed near perfect accuracy for detection of the stimulus, regardless of response type (left hand, right hand, verbal), and well above chance accuracy for indicating location (Pinto et al., 2017a). Even more interesting, however, is that there was no observed interaction between response type and stimulus location (left visual field, right visual field).
October 6, 2025 at 2:35 AM
kh0rish.bsky.social
Because I'm not making 3 billion posts, here ya go. Planning first, and then a little extra
“Planning is the process of thinking regarding the activities required to achieve a desired goal.” - Thinking and activities are the combined outcome of the sensory perception-motility-memory loop. At their most basic, sensations (like appetitive molecules binding to chemosensory receptors or light to photosensory receptors) immediately cause transformation cascades that trigger a movement response. At their most complex, entire organs respond to the environment, passing through the brain (for modulation by memory, emotion, etc) and out to the muscles for action. “Memory” is the many ways to modulate this response, at its most basic, methylating a sensory receptor so it does not immediately trigger the cascade. At the most complex, not acting in the same way you did when experiencing a similar event, because now entire cells, tissues, and organ structures can be changed at far grander scales while still reflecting the simpler act of methylation. The more cells involved, the greater the variety in sensations, possible movements, and possible modulations. - Goals are powered by internal drives, which are the result of biochemical needs. Processes do not work without various biomolecules, and decreases in the concentrations of some (or increases in the concentrations of others) kick compensatory systems on. Breathing hard because you not only need more oxygen but have to expel carbon dioxide. Searching for food because you are hungry. Seeking companionship because you are sad. Playing games because you are bored. Learning something for the thrill and satisfaction of mastery or improving your material conditions. “An important aspect of planning is its relationship to forecasting. Forecasting aims to predict what the future will look like, while planning imagines what the future could look like, and how to get there.” - This comes back to the sensory perception-motility-memory loop and goals. Your memory of your experiences (sensations and actions) help you anticipate… Being a loop, there is no beginning or end, and sorting them out is impossible. Did you think of that thing because a sensation made you think of it, a memory triggered it, or a drive/need? Then after thinking that thing, did you follow thought breadcrumbs to others? Did you ignore it? What is the trigger and what is the response, and which of your tens of trillions of cells kicked it off? With such huge variety, does it matter? It doesn’t, nothing is rote with this many moving parts. All of this comes about from the interactions of molecules, driven by electrostatic forces/chemical bonding affinities that determine the stickiness of things and how often they will be near each other and where they will be located in the cell, steric forces shaping bond angles that determine molecular structures, the elements and their valences determining various other properties, etc. Things just happen. None of this was designed, none of this reflects a platonic solid, there are no instructions (thats not how genes work), molecules don’t read each other or transmit information. Nothing but physical interactions.
October 6, 2025 at 1:20 AM
kh0rish.bsky.social
“Fiance,” thank you
A tiger running away from a little boy while giggle-shouting “woo hoo hoo hoo” after doing the tying-his-shoes-together prank
October 5, 2025 at 4:02 PM
kh0rish.bsky.social
Ayyyyy
bsky.app/profile/kh0r...
However, in Al, our programs to a great degree are problems rather than solutions. If a researcher tries to write an "understanding" program, it isn't because he has thought of a better way of implementing this well-understood task, but because he thinks he can come closer to writing the first implementation. If he calls the main loop of his program "UNDERSTAND", he is (until proven innocent) merely begging the question. He may mislead a lot of people, most prominently himself, and enrage a lot of others. What he should do instead is refer to this main loop as "GO034", and see if he can convince himself or anyone else that G0034 implements some part of understanding. Or he could give it. a name that reveals its intrinsic properties, like NODE-NET-INTERSECTION-FINDER, it being the substance of his theory that finding intersections in networks of nodes constitutes understanding. If Quillian <1969> had called his program the "Teachable Language Node Net Intersection Finder", he would have saved us some reading. (Except for those of us fanatic about finding the part on teachability.)
October 4, 2025 at 5:11 AM
kh0rish.bsky.social
We always lose him
October 3, 2025 at 9:41 PM
kh0rish.bsky.social
I am endlessly amused that despite the immense amount of shit I talk about AI, robots, mind uploads etc, and being absolutely positive that none of it is happening ever…

I do love me some scifi about it. Rewatching pantheon right now with thoughts like “all of this is fn absurd” but my face like:
Zendaya laying on her side grinning from ear to ear as she stares at her cellphone
October 3, 2025 at 8:21 PM
kh0rish.bsky.social
I gotta stop going on this goddamn website but its an unlimited source of nonsense I might have to deal with in the development of my model.

Anyway, the answer to these ridiculous questions is...these assertions arent how biology works it at all. But nooo we cant consider that possibility can we?
intalentive: Generative world models seem to be doing ok. Dreamer V4 looks promising. I’m not 100% sold on the necessity of EBMs. Also I’m skeptical that self-supervised learning is sufficient for human level learning. Some of our ability is innate. I don’t believe it’s possible for statistical methods to learn language from raw audiovisual data the way children can. Human DNA has under 1GB of information content in it. Most of which isn't even used in the brain. And the brain doesn't have a mechanism to read data out from the DNA efficiently. This puts a severe limit on how much "innate knowledge" a human can possibly have. Sure, human brain has a strong inductive bias. It also has a developmental plan, and it follows that plan. It guides its own learning, and ends up being better at self-supervised learning than even the very best of our AIs. But that guidance, that sequencing and that bias must all be created by the rules encoded in the DNA, and there's only this much data in the DNA. ACCount37: It's quite possible that the human brain has a bunch of simple and clever learning tricks that, if we pried out and applied to our AIs, would give us x100 the learning rate and x1000 the sample efficiency. Or it could be that a single neuron in the human brain is worth 10000 neurons in an artificial neural network, and thus, the biggest part of the "secret" of human brain is just that it's hilariously overparameterized. intalentive: The complexity of the human body surely weighs in at over 1 GB. I think of DNA analogously to the rules of cellular automata. The entropy of the rules is much less than the entropy of the dynamical system the rules describe. The body is filled with innate knowledge. The organs all know what to do. The immune system learns to detect intruders (without synapses). Even a single cell organism is capable of complex and fluid goal-oriented behavior, as Michael Levine attests.
October 3, 2025 at 6:41 PM
kh0rish.bsky.social
They stole this girls muhfuckin face this WHOLE industry is scams and fraud from dork ass losers that loathe the concept of consent, kill all of it with fire until we get rid of these companies istg
October 2, 2025 at 4:46 AM
kh0rish.bsky.social
Al pacino disagrees
A man sitting in front of two computer monitors talking to an AI blonde woman in the movie S1m0ne
October 2, 2025 at 4:05 AM
kh0rish.bsky.social
For seven. Minutes.
October 1, 2025 at 7:48 AM
kh0rish.bsky.social
Oh man I really liked the first two Saw movies but it kinda fell off

Wait a damn minute
Laura loomer shopped as the creppy puppet from the Saw films. Red spirals on the protruding cheeks. Pale pale skin. Fuckass shaggy black bob
October 1, 2025 at 12:08 AM
kh0rish.bsky.social
I also need everyone to chill out with the "human nature!" stuff too. Most of the terrible things people bring up are actually results of emotional immaturity and trauma, not our "nature." We absolutely don't have to live like this or endure it forever.
If it is held that man is "by nature" an uninventive tribesman and an inventive businessman, a submissive slave and a proud craftsman an independent hunter and a dependent wage-worker, then either man's "nature" is an empty concept, or man's "nature" depends on material and historical conditions, and is in fact a response to those conditions.
September 30, 2025 at 3:18 PM
kh0rish.bsky.social
A model of one such system at work:
September 30, 2025 at 12:35 PM
kh0rish.bsky.social
You see this? Its a molecular hypodermic needle some bacteria species use to inject DNA or other weaponry. You think we have or can build anything like this?
Type IV secretion * Type IV secretion systems (T4SSs) * are unique amongst the systems characterised to date in that they can translocate proteins both in a one-step and two-step manner.
Furthermore, T4SSs are not just limited to the transport of proteins; they can also function to transport DNA. * Once of the best studied T4SSs is that of Agrobacterium tumefaciens, which utilises a T4SS to export transfer-DNA (t-DNA) into dicotyledonous plants, where the single stranded DNA which is transferred contains oncogenes, resulting in tumour formation in the plant. 3D model of the type 4 secretion system components. Looks like a carousel of doom stacked at the top of the Eiffel tower
September 30, 2025 at 12:35 PM