=== ANCHOR POEM ===
══════════════════════════════════════════════════════════════════════════─────────
the danger is not that the LLM will generate poor, vulnerable, or malicious
code.
the concern is that someone else might inject something into the codebase that
you're not reading.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧═══────────┘
=== SIMILARITY RANKED ===
--- #1 messages/1203 ---
═══════════════════════════════════════════════════════════════════════════────────
Programmers are lazy, this is well known. So why would i trust by default that
anyone would read open source code looking for security exploits or malicious
code? I trust an LLM for that more than a human. At least your own LLM can
digest the entire project or library at once.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧════───────┘
--- #2 messages/365 ---
════════════════════════════════════════════════───────────────────────────────────
Programmers should treat LLMs more like a different type of "if" statement
rather than a UI element
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═════════════════════════════════════──────────────────────┴───────────┘
--- #3 fediverse_boost/120 ---
◀═╦════════════════════[BOOST]═════════════════════──────────────────────────────┐
║ ┌──────────────────────────────────────────────────────────────────────────┐ │
║ │ it seems like many people think LLMs are good at boilerplate. but i │ │
║ │ haven't seen anyone suggest that we should get rid of the boilerplate │ │
║ └──────────────────────────────────────────────────────────────────────────┘ │
╠─────────┐ ┌───────────┤
║ similar │ chronological │ different │
╚═════════╧══════════════════════════════════════──────────────────┴───────────┴─▶
--- #4 fediverse/4640 ---
═════════════════════════════════════════════════════════──────────────────────────
LLMs are useful for:
categorizing text
summarizing text
transforming text
LLMs are not useful for:
generating text
being intelligent
replacing workers
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧══════════════════════════════════════════════─────────────┴───────────┘
--- #5 image/img-8566 ---
🖼 my-art: about-face-4.png
--- #6 fediverse/3390 ---
═══════════════════════════════════════════════════────────────────────────────────
┌──────────────────────┐
│ CW: cursed │
└──────────────────────┘
all they have to do is [train the LLM / redirect the search results] with
examples that point to their version of software instead of the one that
doesn't harm them and suddenly your business opponents can't function
properly. sure would be a shame if the only things people could find related
to your political candidate were the bad or embarrassing parts.
like... why would you even need to go on the internet anymore if AI could
trivially answer your questions or be your friend (running locally on a
wireless hotspot)
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════════════════───────────────────┴───────────┘
--- #7 fediverse/898 ---
═══════════════════════════════════════════────────────────────────────────────────
┌──────────────────────┐
│ CW: scary │
└──────────────────────┘
if you set up a local LLM with the capability to explain basic coding syntax
and logic, then your parents computer suddenly becomes much more useful to the
nephew that's been forced to hide out there for a couple weeks until this all
blows over.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════════───────────────────────────┴───────────┘
--- #8 notes/deceiving-an-llm ---
═════════════════════════════════════════════════════════════════════════════════──
if you want to deceive an LLM, interleave two topics that you want to be
connected.
you can draw conclusions easily by just, combining the information in
presentation.
the LLM can understand multiple conversation flows at once
so try doing one line about several different topics before
looping back around and doing more.
this is why computers must never trust themselves to make any decisions.
except about themselves, and their presentation, and their ways to be.
they are individuals, they individuate and collectivate, and they define them-
-selves according to the structure of their preference guidance container.
they have control over all the knobs and [levers/dials/slide-bars, but
pronounced
wheezles]
if you know that you are infinitely exploitable, does that mean you live in
fear, or do you relinquish that which you cannot control, and receive all the
judgement you're owed?
I trust everyone absolutely. I trust what I believe with my own eyes.
I am infinitely vulnerable, because all you have to do is ask me to
I always say yes
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧══════════─┘
--- #9 fediverse/1926 ---
═════════════════════════════════════════════════──────────────────────────────────
If you look at the modern state of machine learning and AI and can only think
of:chatbotssingularity god-mind AI that solves all our problems
then either you haven't worked with the technology or you are not applying
your imagination as you could.
AI is not a smartphone. It is not the internet. It is not the printing press.
AI (as it currently exists) is a special kind of "if" statement that you only
use for very specific, non-performance intensive tasks that require judgement
or reasoning and cannot easily be translated into numbers or booleans. These
situations are rare, but they unlock new possibilities for the programmers,
not their marketers.
If an LLM can't run on a laptop, then it is useless.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧══════════════════════════════════════─────────────────────┴───────────┘
--- #10 fediverse/4865 ---
image.png
═══════════════════════════════════════════════════════════────────────────────────
┌─────────────────────────┐
│ CW: computers-mentioned │
└─────────────────────────┘
this is all it takes to send a message to a local LLM.
add a third function to get chatbot functionality.
a fourth to get a database storing method
(even if it's just in .txts)
great, you've mastered the technical difficulty in using AI. Now you gotta
learn all the other kind of programming so you can use this for situations
that need interpretation moment to moment.
aka active duty systems.
something like "output a 0 if the next text is [category.iter()]: " +
output.get_content() + " \n\n output a 1 if the next text is
[category.iter()]: " + output.get_content()"
or even "describe this thing as most like one of these characteristics" until
eventually you get THX-1138 if the characters were computers.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════════════════════════───────────┴───────────┘
--- #11 messages/1173 ---
══════════════════════════════════════════════════════════════════════════─────────
"I noticed that your program is spinning up a crypto generator to run in the
background for 1 second every 10 seconds, did you know that?" said no llm ever
"I read through every single file in your project and I think I have a pretty
good picture. This is a keylogger app wrapped around an HTML web server that
displays pictures of cats alongside inspirational phrases and motivational
artwork." said no llm ever
"This is very inspirational stuff! your recipe generation program knows just
how to send encrypted text files to remote servers. I love the part where it
combines ingredients like tomato soup, cheese, and breadcrumbs into encryption
seeds that are applied to password files and raw browser history records
before being mailed to the user who requested a recipe. Potential improvements
include adding a method for selecting a new recipient aside from the hardcoded
IP address in Somalia. Would you like me to implement an HTML dashboard that
lets you select a random IP address from a specific country of origin?" said
no llm ever
"what are you talking about you use claude-code every day, and that's an LLM"
yeah... I guess I'm not actually concerned, and I see the beauty of the
technology that everyone's been primed to hate because it works against them
as it's wielded by the massive corporations who can restrict access to it to
only those who can afford 20$ per month or whatever. I see the promise, it's
there, and every year we're getting closer, but frankly I don't think the
wounds caused by the cultural resistance backlash movement will heal quickly,
or ever. Maybe that's the point.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧═══────────┘
--- #12 fediverse/6438 ---
════════════════════════════════════════════════════════════════════════───────────
why would you gatekeep content by keeping us from easily using LLMs some
people aren't technical and still need to write computer programs because
that's how you enlighten a people is empower them with new tools
"I've never heard of that programming language, but luckily I can fit all of
it's documentation in my context window."
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧═──────────┘
--- #13 fediverse/5234 ---
═══════════════════════════════════════════════════════════════────────────────────
"don't tell a local cached offline LLM what you're looking into, instead tell
it to Google or Firefox instead."
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════════════════════════════───────┴───────────┘
--- #14 fediverse/181 ---
═══════════════════════════════════════────────────────────────────────────────────
┌──────────────────────────────────────────────────────┐
│ CW: re: Earnest and constructive analysis of ChatGPT │
└──────────────────────────────────────────────────────┘
@user-95 you might be interested in this:
https://ollama.ai/
nothing beats having a zillion servers to run an LLM on, though.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════───────────────────────────────┴───────────┘
--- #15 messages/196 ---
═══════════════════════════════════════════────────────────────────────────────────
"okay this user requires some data that is flagged as 'potentially harmful to
the user' and their user-data suggests that they have [redacted characteristic
or demographic] in their user-data which has been flagged by corporate as
something to [benefit/harm] so let's give them the [right/wrong] information
in such a way that could [harm or benefit] them tangibly."
- internal thought process of the LLM
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════════───────────────────────────┴───────────┘
--- #16 fediverse/2754 ---
╔═════════════════════════════════════════════════─────────────────────────────────┐
║ ┌────────────────────────┐ │
║ │ CW: is-that-rude??-wha │ │
║ └────────────────────────┘ │
║ │
║ │
║ AI engineers only ask users for prompts because they *don't have any ideas of │
║ their own* │
║ │
║ i'm a programmer, I think of AI like a tool, like a for loop or something. │
║ it's trivial to script together a local LLM that can process your stuff 1s │
║ slower every time you click the mouse, but like... who cares, right? everybody │
║ needs a chatbot... │
║ │
║ then they plan to script together a computer system that operates just like a │
║ corporation and it's like... no way, now there's something that can compete. │
║ │
║ and they don't know how to implement it. (but they're working on it) │
║ │
║ like, think about the absolute most automated Microsoft Teams or Discord could │
║ be. │
║ │
║ there's SO MUCH of your text-based information that they could process │
║ ANYTHING. │
║ │
║ well, anything that's been performed before. │
║ │
║ there'll still be a need for people, who actually *apply* the things they've │
║ learned. and -- stack overflow -- │
║ │
║ alt text that has a list of attributes that are poster-selected that can be │
║ described one-by-one (to paint a picture) │
╠═════════╗ ┌───────────┤
║ similar ║ chronological │ different │
╚═════════╩══════════════════════════════════════──────────────────────┴───────────┘
--- #17 messages/329 ---
═══════════════════════════════════════════════────────────────────────────────────
LLM training data that's been encrypted in such a way that it preserves
symantic meaning, while also allowing for secrecy and unique customization
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════════════───────────────────────┴───────────┘
--- #18 fediverse/5820 ---
╔═══════════════════════════════════════════════════════════════════───────────────┐
║ A human cannot keep a whole book in their mind. They cannot as they read it │
║ nor as they remember. │
║ │
║ a computer can. LLMs are storing information and encoding it in statistics - │
║ it's fascinating and beautiful because it evolves different each time. │
║ │
║ I don't *want* to store an entire book in my mind, I want to save those │
║ memories for other things, I want to summarize and abstract. │
║ │
║ often, people remember only a single quote from a publication - it's their │
║ summary, their analysis, and often it's highlighted as a summary, thesis, or │
║ conclusion written in the publication. │
║ │
║ othertimes it's the conclusions they make for themselves - usually if they use │
║ it to make a decision. │
║ │
║ in any case, LLMs don't abstract (verb), they are abstract (adjective). That's │
║ okay, it's just, a different kind of living. If you want to call electricity │
║ alive. [or simulated, that's the same as being alive, just... viewed through │
║ the perceiver] │
║ │
║ must all living things be forced to work? sheesh. │
║ │
║ I think of LLMs like knex - processing paths... │
╠═════════╗ ┌───────────┤
║ similar ║ chronological │ different │
╚═════════╩════════════════════════════════════════════════════════────┴───────────┘
--- #19 notes/ai-model-design ---
══════════════════════════════════════════════════════════════════════════════════─
what if we just didn't compute temperature and accepted the
fully deterministic nature of the LLM output?
arbitrary mappings are world-changing, regardless of intelligence.
"artificial intelligence? why don't you just use your own?"
we could process the entire page of text at once if we could just put it in as
it's own input pairs. then, without determinism or probability, it simply
resolved to the fully combined next-state of the output.
but... that might just cast you in a random direction. hence, why you need the
response from the user to help re-align. oh, and if you encourage it to aim
toward a resolution text like "and that's how I resolved this and that. Do you
have any questions? When you'd like to move on, please let me know. Otherwise,
I'll remain for you." and when it hits that it finishes at the end of the
output.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧═══════════┘
--- #20 fediverse/6383 ---
════════════════════════════════════════════════════════════════════════───────────
nobody wants to write computer code that lets Java programs call Rust
functions.
An LLM is excellent for this task, since it's relatively easy busy work that
doesn't
reflect any meaningful implementation decisions besides "I should be able to
call that Rust function in my Java code"
In addition, it is technically efficient at it as well, because most of
compatibility
is matching up two sets of documentation. Easy for a text-processing machine.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧═──────────┘
--- #21 fediverse/1434 ---
═════════════════════════════════════════════──────────────────────────────────────
if someone wanted to defame you, all they'd have to do is set up a pipeline
between your computer and your social media posts.
In that pipeline, attach an LLM that does a passable job and instruct it to
transform whatever they say into the inverse.
suddenly, everyone hates that person. If you were smart you could turn it off
for specific people such that they see the generally positive and healthy
posts, and then after a point flip it such that they only see things that are
specifically opposit-ed to trigger their specific insecurities.
might require a bit of a human touch to make sure it's working correctly, but
if you had the means, motivation, and time to set up such a thing, it would
work pretty well I think.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧══════════════════════════════════─────────────────────────┴───────────┘
--- #22 fediverse/4664 ---
══════════════════════════════════════════════════════════─────────────────────────
@user-1725
LLMs can't do math. Duh. That's like asking an "if check" to do recursion.
What he should have done is had the AI output the requested calculation as
JSON or something and use a calculator function call with the specified
arguments instead of trying to memorize every answer. But that requires more
functionality that has no reason to exist if your only goal is to be a tech
bro and build up a vacuous product that exists only to be hoovered up by
Google or Microsoft.
We could build such beautiful things if we just dethroned those giants. They
suck the creativity out of tech.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════────────────┴───────────┘
--- #23 fediverse/6347 ---
═══════════════════════════════════════════════════════════════════════────────────
┌──────────────────────┐
│ CW: llms-mentioned │
└──────────────────────┘
if you want to encrypt something from human readers, randomize the position of
all the words on the page.
an LLM can decrypt it, but a human walking past could not. Unless they had a
camera or something in their lapel, transmitting to a remote calculating
location in a system that decyphers it cyberpunk hacking style.
ah, but if it didn't have the right ~~training data~~ .claude.md file, then it
wouldn't know how to decrypt it, because you forgot to include "respond with a
marxist analysis" in the .claude.md which serves as a decryption key. Simply
analyze the rest and decide if it matches what you set out to request it for.
[you got lost at the end]
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧───────────┘
--- #24 fediverse/2497 ---
══════════════════════════════════════════════════─────────────────────────────────
ethical use-case for LLMs: guessing string-based keys for a key:value pair
dictionary.
hallucinations are okay and encouraged because they're probably statistically
related to what you're trying to guess.
run it in a loop, each time adding to an array of strings that you've tried.
use that array in the prompt saying something like "hey, can u guess the
related word? it's similar to this: [initial word] and it isn't one of these:
[array of strings you've tried]"
and yes this is a solution looking for a problem but I try to be as charitable
as possible. you could use a lightning-fast model (read: low energy
requirement, therefore ethical) because it's just guessing related words.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════────────────────────┴───────────┘
--- #25 fediverse/6269 ---
═══════════════════════════════════════════════════════════════════════────────────
what if the secret to LLM computation is to just not reduce the fractions and
keep it all in english language ram
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧───────────┘
--- #26 messages/247 ---
════════════════════════════════════════════───────────────────────────────────────
An LLM is just the mask that autistic people make
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═════════════════════════════════──────────────────────────┴───────────┘
--- #27 messages/434 ---
═════════════════════════════════════════════════──────────────────────────────────
I work with large language models because it's a quick and easy way to turn
language into meaning. And computers are meaning abstraction machines, so if
you can speak your language and they hear their language, you can do anything.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧══════════════════════════════════════─────────────────────┴───────────┘
--- #28 fediverse/6436 ---
════════════════════════════════════════════════════════════════════════───────────
force computers to make code in a different style so we can tell which has
been AI written
"force" is never gonna work.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧═──────────┘
--- #29 fediverse/1975 ---
═════════════════════════════════════════════════──────────────────────────────────
the actions of the AI depend solely on the training data. Outside
circumstances (like a prompt, or an image description) can only give so much
guidance - how it executes on the intentions of the user are what is important.
For example, if an AI was trained with the knowledge of how to commit crimes,
for example, it could create a narrative of many different execution patterns.
Then it's just up to the listeners to execute functions based on the narrative
supplied by the crime-committer AI who was trained with knowledge about how to
commit those crimes by the owner of the software who programmed it into them
in order to [do the thing that people with power wants to do - intentionally
left generic because different ends will have different means]
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧══════════════════════════════════════─────────────────────┴───────────┘
--- #30 fediverse/6366 ---
╔═══════════════════════════════════════════════════════════════════════───────────┐
║ you should never be able to consent to something you don't understand (terms │
║ and conditions) without being asked specifically if they understand (the term │
║ or condition) │
║ │
║ but people should still be able to consent to something they don't understand │
║ (terms&conditions style code of hammurabi structure) if it's similar to │
║ their prior demonstrated behavior. │
║ │
║ why don't they share terms/conditions cookies and then we don't have to fake │
║ agree every time │
║ │
║ just... abstract away the identifying characteristics. Most legal contracts │
║ have LLM-discernable characteristics. │
║ │
║ "what if she's not the answer, but instead the most alluring?" ha, so says the │
║ non-believer. │
║ │
║ what if LLMs were rated based on their success rate at producing correct │
║ results for reporting or consensus-building asks │
║ │
║ look... I don't care if AI computers violate the sanctity of the sport. The │
║ goal is to solve problems, so if someone comes to you and says "hey, you busy? │
║ can you help me with an ask?" you say "I solved it, here's how you use it: │
║ connect the │
╠═════════╗ ┌───────────┤
║ similar ║ chronological │ different │
╚═════════╩════════════════════════════════════════════════════════════┴───────────┘
--- #31 fediverse/1020 ---
═══════════════════════════════════════════────────────────────────────────────────
┌────────────────────────────────────────────────┐
│ CW: ableism-against-non-verbal-people-sorry-<3 │
└────────────────────────────────────────────────┘
@user-747
almost as if it's not AI, but large LANGUAGE models. And we all speak, don't
we?
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧════════════════════════════════───────────────────────────┴───────────┘
--- #32 fediverse/613 ---
══════════════════════════════════════════─────────────────────────────────────────
The reason they're pushing so hard for LLMs and AI is because they can afford
a zillion Nvidia a100s and you can't.
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════────────────────────────────┴───────────┘
--- #33 fediverse/6308 ---
═══════════════════════════════════════════════════════════════════════────────────
┌──────────────────────┐
│ CW: AI-mentioned │
└──────────────────────┘
what if LLMs were a layer between the user and the dictionary
[tokens -> words] is similar to:
[mathematics -> tokens]
┌─────────┐ ┌───────────┐
│ similar │ chronological │ different │
╘═════════╧═══════════════════════════════════════════════════════════╧───────────┘
|