"my personality can be reproduced probabilistically" is not the flex people think it is
creepy tech bro demands even creepier behavior from already incredibly creepy tech bro and his product
April 13, 2024 at 4:08 PM
"my personality can be reproduced probabilistically" is not the flex people think it is
Chatbots — LLMs — do not know facts and are not designed to be able to accurately answer factual questions. They are designed to find and mimic patterns of words, probabilistically. When they’re “right” it’s because correct things are often written down, so those patterns are frequent. That’s all.
June 19, 2025 at 11:21 AM
Chatbots — LLMs — do not know facts and are not designed to be able to accurately answer factual questions. They are designed to find and mimic patterns of words, probabilistically. When they’re “right” it’s because correct things are often written down, so those patterns are frequent. That’s all.
Generative AI is essentially copy paste but with large datasets analyzed probabilistically. Maybe that's useful but it's not intelligence and doesn't justify the absurd amounts of investment it's attracted
January 28, 2025 at 1:07 AM
Generative AI is essentially copy paste but with large datasets analyzed probabilistically. Maybe that's useful but it's not intelligence and doesn't justify the absurd amounts of investment it's attracted
That's hilarious, its almost like deterministic computing is cheaper and more accurate than doing the same thing probabilistically. Don't tell management
March 17, 2026 at 3:50 PM
That's hilarious, its almost like deterministic computing is cheaper and more accurate than doing the same thing probabilistically. Don't tell management
the program is tokenizing inputs into vectors, then doing matrix multiplication to probabilistically compute which token should follow those tokens.
as usual, you are mystified by systems beyond your understanding and come to a bad conclusion (coincidentally the one you know will get you yelled at)
as usual, you are mystified by systems beyond your understanding and come to a bad conclusion (coincidentally the one you know will get you yelled at)
August 9, 2025 at 4:43 PM
the program is tokenizing inputs into vectors, then doing matrix multiplication to probabilistically compute which token should follow those tokens.
as usual, you are mystified by systems beyond your understanding and come to a bad conclusion (coincidentally the one you know will get you yelled at)
as usual, you are mystified by systems beyond your understanding and come to a bad conclusion (coincidentally the one you know will get you yelled at)
i mean probabilistically at least some liberal centrists must really be actual liberal centrists and not secret cryptofascists
I kinda like the stance that's come out since Jan 20
February 4, 2025 at 12:15 AM
i mean probabilistically at least some liberal centrists must really be actual liberal centrists and not secret cryptofascists
there are seven billion real people in this world and you choose to interact with the machine that spits out probabilistically sewn together bits of other people's conversations?
February 7, 2026 at 7:07 AM
there are seven billion real people in this world and you choose to interact with the machine that spits out probabilistically sewn together bits of other people's conversations?
People have a hard time thinking probabilistically.
Sometimes things that give you a worse chance to win STILL LEAD TO WINS.
Sometimes things that give you a worse chance to win STILL LEAD TO WINS.
It won them the game.
November 23, 2025 at 9:31 PM
People have a hard time thinking probabilistically.
Sometimes things that give you a worse chance to win STILL LEAD TO WINS.
Sometimes things that give you a worse chance to win STILL LEAD TO WINS.
The advice is “Think for yourself and do your own research”?
And their plan to do this is “ask the chatbot to give you the probabilistically most average answer w a little random error thrown in”?
Do I have that right?
And their plan to do this is “ask the chatbot to give you the probabilistically most average answer w a little random error thrown in”?
Do I have that right?
August 24, 2026 at 12:44 PM
The advice is “Think for yourself and do your own research”?
And their plan to do this is “ask the chatbot to give you the probabilistically most average answer w a little random error thrown in”?
Do I have that right?
And their plan to do this is “ask the chatbot to give you the probabilistically most average answer w a little random error thrown in”?
Do I have that right?
“Your conception of gender is like the Bohr model atom where genders exist and move between fixed energy level shells, whereas Chris’s conception of gender is like the quantum model atom where genders probabilistically cluster into orbital clouds”
@turn3coco.bsky.social, 2025
@turn3coco.bsky.social, 2025
November 25, 2025 at 4:46 PM
“Your conception of gender is like the Bohr model atom where genders exist and move between fixed energy level shells, whereas Chris’s conception of gender is like the quantum model atom where genders probabilistically cluster into orbital clouds”
@turn3coco.bsky.social, 2025
@turn3coco.bsky.social, 2025
If the threshold for being alive is probabilistically returning answer-shaped text when prompted, a magic eightball is alive
September 4, 2026 at 3:46 PM
If the threshold for being alive is probabilistically returning answer-shaped text when prompted, a magic eightball is alive
a new york times author is a stochastic parrot that probabilistically calls large language models stochastic parrots
December 10, 2025 at 3:32 PM
a new york times author is a stochastic parrot that probabilistically calls large language models stochastic parrots
"Can act like it does" would make sense as it is already probabilistically "acting" like a person. That doesn't mean it "feels pain" just that the human words the computer assembles came from people who felt pain
September 26, 2026 at 4:41 PM
"Can act like it does" would make sense as it is already probabilistically "acting" like a person. That doesn't mean it "feels pain" just that the human words the computer assembles came from people who felt pain
probabilistically, i dont give a fuck
October 6, 2025 at 11:57 AM
probabilistically, i dont give a fuck
Underappreciated thread of the AI discussion today is the perspective that you can’t build complex, predictable systems from things that probabilistically fail.
Speaking generally, that’s not how anything works. *Everything* fails. Yet we can still make reliable complex systems.
Speaking generally, that’s not how anything works. *Everything* fails. Yet we can still make reliable complex systems.
September 6, 2026 at 9:35 AM
Underappreciated thread of the AI discussion today is the perspective that you can’t build complex, predictable systems from things that probabilistically fail.
Speaking generally, that’s not how anything works. *Everything* fails. Yet we can still make reliable complex systems.
Speaking generally, that’s not how anything works. *Everything* fails. Yet we can still make reliable complex systems.
The most disappointing thing I’ve found using AI agents is you can’t enforce guardrails via memory or CLAUDE .md files.
To an LLM, your instructions are just more text to probabilistically reason about.
“Don't do X” simply cannot work. You must enforce deterministic rules outside the system.
To an LLM, your instructions are just more text to probabilistically reason about.
“Don't do X” simply cannot work. You must enforce deterministic rules outside the system.
June 28, 2026 at 4:40 AM
The most disappointing thing I’ve found using AI agents is you can’t enforce guardrails via memory or CLAUDE .md files.
To an LLM, your instructions are just more text to probabilistically reason about.
“Don't do X” simply cannot work. You must enforce deterministic rules outside the system.
To an LLM, your instructions are just more text to probabilistically reason about.
“Don't do X” simply cannot work. You must enforce deterministic rules outside the system.
regulating vehicles that behave probabilistically and are operated by a company whose unit economics depend on minimizing human oversight of the fleet is going to be... an interesting governmental challenge
July 17, 2026 at 12:01 AM
regulating vehicles that behave probabilistically and are operated by a company whose unit economics depend on minimizing human oversight of the fleet is going to be... an interesting governmental challenge
You drive into the parking stew. Your car is absorbed into the glittering, molten mass—later you will reclaim a strange vehicle, probabilistically reconstituted
May 12, 2026 at 3:53 AM
You drive into the parking stew. Your car is absorbed into the glittering, molten mass—later you will reclaim a strange vehicle, probabilistically reconstituted
starting to think most people do not think probabilistically or something
May 31, 2026 at 3:09 PM
starting to think most people do not think probabilistically or something
Gen. AI is unprofitable, unsustainable, and fundamentally limited in what it can do thanks to the fact that it's probabilistically generating an answer.
It's not the future - it's a regurgitation of the past.
wheresyoured.at/pop-culture/
It's not the future - it's a regurgitation of the past.
wheresyoured.at/pop-culture/
July 8, 2024 at 5:49 PM
Gen. AI is unprofitable, unsustainable, and fundamentally limited in what it can do thanks to the fact that it's probabilistically generating an answer.
It's not the future - it's a regurgitation of the past.
wheresyoured.at/pop-culture/
It's not the future - it's a regurgitation of the past.
wheresyoured.at/pop-culture/
There's a phrase a variant of which I use often, in many contexts:
This was not deterministically inevitable, but it was probabilistically predictable.
This was not deterministically inevitable, but it was probabilistically predictable.
August 28, 2023 at 3:42 PM
There's a phrase a variant of which I use often, in many contexts:
This was not deterministically inevitable, but it was probabilistically predictable.
This was not deterministically inevitable, but it was probabilistically predictable.
Scientific illustration is superior to photography for many situations, because an artist can subtly emphasize key features & "standardize" the anatomy of an organism where photography can only capture individuals. AI can't do this, because it's still probabilistically guessing what goes where
June 5, 2026 at 5:14 PM
Scientific illustration is superior to photography for many situations, because an artist can subtly emphasize key features & "standardize" the anatomy of an organism where photography can only capture individuals. AI can't do this, because it's still probabilistically guessing what goes where
When a chatbot arranges words into a format of an apology or admission of guilt it does so because it was instructed to do so by the user or because that is what the statistical model indicates is the words that should be probabilistically used in that situation.
Not “self reflection”
Not “self reflection”
July 21, 2025 at 1:20 AM
When a chatbot arranges words into a format of an apology or admission of guilt it does so because it was instructed to do so by the user or because that is what the statistical model indicates is the words that should be probabilistically used in that situation.
Not “self reflection”
Not “self reflection”
Yeah, being chill is probabilistically your best way of getting out of a lot of potential jams…
April 26, 2026 at 2:35 PM
Yeah, being chill is probabilistically your best way of getting out of a lot of potential jams…
It's frustrating how many news stories these days are journalists being impressed that a chatbot trained on the public source code of the entire internet and every exam question can probabilistically generate solutions to doing stuff with computers better than someone who is bad at computers.
July 22, 2026 at 12:16 PM
It's frustrating how many news stories these days are journalists being impressed that a chatbot trained on the public source code of the entire internet and every exam question can probabilistically generate solutions to doing stuff with computers better than someone who is bad at computers.