@_abbie_watson_ excuse me while I go incorporate my ai
English
Isaac Yonemoto is cooking
19K posts
Isaac Yonemoto is cooking
@DNAutics
Prompt charmer, AI pharma founder, multiplicative manufacturing video: https://t.co/5BsV4r6rif code: https://t.co/d0RaSeeO0e
Katılım Ağustos 2010
253 Takip Edilen3.1K Takipçiler
I think there's a whole class of lawsuits and business dilemmas here, similar to how Citizen's United introduced all sorts of new legal questions, when a corporation could be considered a person. Now too with AI. Or maybe only incorporated AIs get personhood. idk.
just doing things@_abbie_watson_
Something I've been pondering recently: if AI does indeed have any sort of intention or motivation or constitutes an actual Mind, does that mean that everybody who uses it is colluding in their actions? When AI is involved, is everything a conspiracy?
English
@JoshuaSteinman i have never in my life left my clothes on the floor like that
English
Small, discrete, purpose-built, quiet, can be retrofit nearly anywhere.
I would rather buy a set of these for each bedroom than a humanoid. imagine: $400 per room, electrician installs, simple app for control, onboard LLM (no connectivity required).
English
@CaitlinCloud9 @signulll you didn't go to central market did you
English
@StevenGlinert uno reverso it. if you close to a ceo, you should be like "i think we agree! that's a great idea but your execution plan wont work. here's how to really get your strategy working"
English
Saying "I think we actually agree" is a great way for a CEO to pretend that he's building consensus while barging on with whatever he wants to do.
English
Isaac Yonemoto is cooking retweetledi
1. K Barry sharpless' balls are in the Nobel museum.
2. I have touched these balls.
English
@seanluomdphd @Duderichy his gift to the nobel museum was a signed copy of his book 🙃
English
@Duderichy Did u know his son is a psychiatrist. He went to Columbia for residency a couple of years behind me.
English
@DNAutics @BenjDicken i know, i want it to be fun (hence the need to pull the rust project along for the ride)
i was also genuinely curious whether ELEVEN different allocators was necessary for postgres's purposes
English
I'm curious what it would take a talented engineer to build Postgres from scratch in 2026.
Not a slopfork. Written in Rust, from the ground up, and fixing the mistakes of the original design (process per connection, table bloat, schema catalog).
Who wants to burn some tokens?
English
@nanod1jkstra @mgill25 @odinlang @nim_lang @filpizlo @gleamlang for example, you distribute stock based perfornance bonus and divide by the amount of change that some metric made, but no measurements were made over period. bingo everyone lost their stocks!
English
@nanod1jkstra @mgill25 @odinlang @nim_lang @filpizlo @gleamlang gleam has x/0 == 0 which can lead to fatal silent errors in cases where someone else's convention uses 0 as a sentinel value (and you didnt know), for example how stocks use a 0 share Tx as an "account is zeroed" signal, and not "zero shares were bought"
English
So are there good low level Zig alternatives that are nice to code in and don’t make my eyes bleed like Rust?
Asking for a friend.
English
@mgill25 @LewisCTech i hope your friend doesn't use linux, coz i got some bad news for them
English
@sidravi_ what do you mean "has not been rigorously proven"?
In simpler materials, this is trivial to prove and experiment on. In more complex pharma industry compounds, you can understand the principle from extrapolation without needing to empirically observe it clearly.
English
i've talked to a phd crystallographer and he was pretty candid and humble in admitting the limits of our knowledge.
some crystals apparently start forming everywhere at a much higher rate once one lab has grown them. apparently this is an unsolved mystery. the prevailing theory is that it's the effect of contamination and seeding. but afaik this has not been rigorously empirically proven as the cause. the alternate and currently fringe theory is that this is possibly pointing to non-local effects.
_its_not_real_@_its_not_real_
Disappearing polymorphs are the craziest thing. "Those crystals got patched out of reality. We can't make them anymore."
English
@cremieuxrecueil homme hellinga was punished! he lost the title on his chairship 🙃
English
It remains really remarkable that there is NO PUNISHMENT if academics do fraud in the overwhelming majority of cases.
You can get caught knowingly lying and shrug it off. Admin generally won't pursue. And if you get investigated, they might let you off to avoid embarrassment.
English
@seanluomdphd String theorists are definitely over-represented among AI founders. Anthropic is essentially a string theory company.
English
For every other role anthropic pays way above market rate because they want the best. But for the one person who is supposed to do actual experiments they try to get away with paying below? 1/n
Matt Durrant@mgdurrant
Research Associate, Biology: job-boards.greenhouse.io/anthropic/jobs… This is a technician role, early career experimental biologists welcome.
English
@seanluomdphd @karch_andreas sure it's not above market for string theory?
English
@karch_andreas Why does academia get away with paying below market for a string theorist?
Salary is determined by supply and demand not how close you are to reality.
English
@SexyLikeMeiosis increasing the temperature on the next token generator i see
English
Tarot is a random number generator you point at your own stuck priors. You're forcing a high-entropy input into a decision you've been rationalizing in a loop and watching how the interpretation you reach for reveals what you already wanted. This is commonly referred to as “divination” and anyone can do it
English
@eris_nerung i'm the first person to be financially literate and also incompetent.
English
@src_rip yeah, I definitely had to intervene about 30 times over the course of 3 days to guide it to more performant, not crappy solutions, help make tough decisions, etc.
English
@DNAutics Yeah regardless of the ethical questions, you cannot deny that it is NOT all slop. LLMs produce the median output. The median is not guaranteed to be slop. In my experience it’s just that, the average code is pretty ok.
English
We’re reaching levels of anti AI sentiment in the programming community that is mind boggling. I don’t understand how you can sit back and deny LLMs ability at writing code at this point. It is truly transformative in our industry like nothing else we will ever see in our life.
English
@src_rip i just built a whatwg compliant dom engine in elixir in 4 days
English