Pillb

@dwarkesh_sp Also if you try to make butterfly wings out of butterflies turns out way less efficient than the naturally developed thing.

There's a quadrillion-dollar question at the heart of AI: Why are humans so much more sample efficient compared to LLM? There are three possible answers: 1. Architecture and hyperparameters (aka transformer vs whatever ‘algo’ cortical columns are implementing) 2. Learning rule (backprop vs whatever brain is doing) 3. Reward function @AdamMarblestone believes the answer is the reward function. ML likes to use pretty simple loss functions, like cross-entropy. These are easy to work with. But they might be too simple for sample-efficient learning. Adam thinks that, in humans, the large number of highly specialised cells in the ‘lizard brain’ might actually be encoding information for sophisticated loss functions, used for ‘training’ in the more sophisticated areas like the cortex and amygdala. Like: the human genome is barely 3 gigabytes (compare that to the TBs of parameters that encode frontier LLM weights). So how can it include all the information necessary to build highly intelligent learners? Well, if the key to sample-efficient learning resides in the loss function, even very complicated loss functions can still be expressed in a couple hundred lines of Python code.

@AsakyGRN Trust me you don’t want virgin animals on the contrary if it turns out one of the animals was unknowingly pregnant you just got extra for free.

Lady cries out after her boyfriend ended their six-year relationship after seeing her bride price list, which included 5 fat cows, 50 broiler chickens, 30 vir&!n goats, 5,100 bags of cement, 500 tubers of yam, ₦10 million for the family, ₦5 million for the community, an iPhone 17 Air for the bride’s mother, iPhone 17 Pro Max for the father, 10 bags of foreign rice, 50 liters of fuel, solar panel, and more. 😭

@AndrewPlaTech This made me itch right in the NDAs

Snover told us he once put a bug back into #PowerShell after already fixing it. The Exchange team had validated against the broken version. Billion 💰 commitments on the line. No time to revalidate. He unfixed it. That's what shipping at scale looks like from the inside.

@Edmontondude But what if the branch is in a cylinder?

@Rainmaker1973 Those trees are going to take you to the Geneva Convention court

One way to stop a wildfire is to start a controlled one early.

@bitcloud This is just a protein tensor

This is the thing "LLMs are conscious" people think is the same as a matrix multiplication.

@vkrajacic A standard language to ask questions to AI? Or queries?

So prompt engineers learned English is imprecise for coding and started using stricter syntax. I wonder how long it will take before they come full circle and land back at programming languages again.

@GothicaAvalon @omgsidewalks A corollary: straighten you back every time you go through a door.

@omgsidewalks Fundamental advice for life: Relax your face and shoulders. Check in as frequently as possible. You’ll be surprised how tense they are a lot of the time! 🌷

I'm 22. Please recommend to me oddly specific life tips. No general “surround yourself with positive people” tips. I want the most random, specific advice possible please.

@Prawcin You penis would have to be a monkey penis for the super AIDS

if you had both a penis and a vagina and you impregnated yourself would your child be a clone or have super aids

Dis the Milky Way pee in the pool?

@tautologer The doom of a lot of geniuses in writing

there is no functional difference between predicting what an intelligent person would say in a situation and actually being intelligent. in order to predict what a smart person would say you need to be at least as smart as them

@Plinz No matter how big your rainstick it won’t make her wet

Google has saddled itself with a "simulated water does not make you wet" philosopher

@realfrothyfriar Invaders will also be tenderized

When you need to tenderize steaks at 6 but need to defend the castle from invaders at 6:30

@TheBakaArts Plaguemiddlemanager

The Plaguemaster - They say he doesn’t create the disease. He decides where it spreads.

@BobGolen Just keep playing, you’ll meet’em

I've never met anyone who's lost at Russian Roulette.

@blacknredtext Is consensus that poop blade sneak slash is the winnig strat right?

Interesting scenario

@jackl80475749 @blacknredtext Sneeze on the blade for good measure

@blacknredtext Option A: Rub your or it’s feces on the knife. Cut it while sleeping, ideally a long slash as opposed to a stab, maximizing odds of infection (don’t blind it cause it’s gonna thrash in response to waking up in pain). Option B: Taint the water supply if able. Wait.

@GreenTextRepost Who sucked the first cow? Who discovered it only worked for female cows?

Anon submission

@QiaochuYuan Where do figs lie in that spectrum? Aren’t there tiny wasps in the figs?

if you decided for ethical reasons not to eat any food whose production requires killing a living organism, animal or plant, what would you be able to eat? i think fruit, berries, nuts, seeds, eggs, milk maybe? which grains can be harvested without killing a plant?

@hetmehtaa Rip to the people that are polite with AI and start with hello

The $20 claude plan lets you to ask Opus 4.7 approximately 3 questions before you hit the limit.