Turn up your output with @friend-dorothy.bsky.social and let the results speak for themselves 💦
💊: friend-of-dorothy.co.uk/pupargon
💊: friend-of-dorothy.co.uk/pupargon
December 21, 2025 at 5:18 PM
Turn up your output with @friend-dorothy.bsky.social and let the results speak for themselves 💦
💊: friend-of-dorothy.co.uk/pupargon
💊: friend-of-dorothy.co.uk/pupargon
suspect a big reason why many academics and others who work in areas where getting facts RIGHT is key are disinterested in using LLMs for research:
they’ve tried it, they keep noticing major errors in output, and they conclude that having to verify all that doesn’t actually save them time.
they’ve tried it, they keep noticing major errors in output, and they conclude that having to verify all that doesn’t actually save them time.
December 21, 2025 at 4:16 PM
suspect a big reason why many academics and others who work in areas where getting facts RIGHT is key are disinterested in using LLMs for research:
they’ve tried it, they keep noticing major errors in output, and they conclude that having to verify all that doesn’t actually save them time.
they’ve tried it, they keep noticing major errors in output, and they conclude that having to verify all that doesn’t actually save them time.
There is such a profound misunderstanding of what the outputs of language models really are: they take in a sequence, a prompt, and output a probabilistically-generated sequence that is the model saying “given the input sequence, here’s the most likely output sequence.”
This isn’t summarization.
This isn’t summarization.
Professional societies keep beclowning themselves buying into a lie about what an LLM "summary" is. They are inherently counterfeit: not an epistemic product of the ideas in the source, but summary-shaped text linguistically based on *other* works (in the training corpus) that use related language.
This is one of the reasons I remain horrified by seeing @historians.org suggest "ways to use gAI" that included this:
December 21, 2025 at 5:05 AM
There is such a profound misunderstanding of what the outputs of language models really are: they take in a sequence, a prompt, and output a probabilistically-generated sequence that is the model saying “given the input sequence, here’s the most likely output sequence.”
This isn’t summarization.
This isn’t summarization.
new business idea. a searchable database of media generated using the most common AI prompts, cutting down on compute spent generating output for the same prompt multiple times. seeking investors
December 21, 2025 at 12:52 AM
new business idea. a searchable database of media generated using the most common AI prompts, cutting down on compute spent generating output for the same prompt multiple times. seeking investors
"Canada should increase output of existing pipeline infrastructure before building new pipelines"
Agree: 72%
Disagree: 29%
Ipsos / Dec 15, 2025 / n=1502 / Online
Agree: 72%
Disagree: 29%
Ipsos / Dec 15, 2025 / n=1502 / Online
December 18, 2025 at 10:43 PM
"Canada should increase output of existing pipeline infrastructure before building new pipelines"
Agree: 72%
Disagree: 29%
Ipsos / Dec 15, 2025 / n=1502 / Online
Agree: 72%
Disagree: 29%
Ipsos / Dec 15, 2025 / n=1502 / Online
In Mr. Beast’s new million dollar challenge they precisely measure a guy’s food intake and feces output and for each gram of nutrients that goes undigested they take away $10,000
December 22, 2025 at 6:17 AM
In Mr. Beast’s new million dollar challenge they precisely measure a guy’s food intake and feces output and for each gram of nutrients that goes undigested they take away $10,000
The rule should not be chosen based on who it excludes.
If the rule means that the output of an entire publishing house, which goes AI-edit forward in 2026, is excluded, the rule is functioning as intended.
If the rule means that the output of an entire publishing house, which goes AI-edit forward in 2026, is excluded, the rule is functioning as intended.
December 20, 2025 at 5:31 PM
The rule should not be chosen based on who it excludes.
If the rule means that the output of an entire publishing house, which goes AI-edit forward in 2026, is excluded, the rule is functioning as intended.
If the rule means that the output of an entire publishing house, which goes AI-edit forward in 2026, is excluded, the rule is functioning as intended.
Say what you will about boomer cultural output, but those racists were cooking when they dropped Deep Purple - Highway Star
December 20, 2025 at 8:33 AM
Say what you will about boomer cultural output, but those racists were cooking when they dropped Deep Purple - Highway Star
Thread: As a translator myself, I echo everything Aliette says. I have also noticed longterm customers suddenly no longer calling, because somewhere there is a machine translation system that is cheaper, even if its output is worse.
Deep breath
Ok I'm going to try and stay calm
But I'll point out that AI translation is literally causing translators to be fired (see the whole Harlequin line in France)
(1/?)
Ok I'm going to try and stay calm
But I'll point out that AI translation is literally causing translators to be fired (see the whole Harlequin line in France)
(1/?)
December 21, 2025 at 6:18 PM
Thread: As a translator myself, I echo everything Aliette says. I have also noticed longterm customers suddenly no longer calling, because somewhere there is a machine translation system that is cheaper, even if its output is worse.
I've seen LLMs change meaning (is to isn't), make up terminology, and if there's numerical information in the source - well, I've seen some *interesting* output.
And that it takes a lot more effort to check the sources against the output than it actually takes to carefully read them yourself.
And that it takes a lot more effort to check the sources against the output than it actually takes to carefully read them yourself.
December 21, 2025 at 4:28 AM
I've seen LLMs change meaning (is to isn't), make up terminology, and if there's numerical information in the source - well, I've seen some *interesting* output.
And that it takes a lot more effort to check the sources against the output than it actually takes to carefully read them yourself.
And that it takes a lot more effort to check the sources against the output than it actually takes to carefully read them yourself.
Oh sorry, one more
I would not ever take the output of any software translation as accurate. Even excluding hallucinations (and that's a big exclude), what you're not getting from it is context dependent meanings (like homophones, idioms, slang, etc). Fastest way to accidentally insult someone
I would not ever take the output of any software translation as accurate. Even excluding hallucinations (and that's a big exclude), what you're not getting from it is context dependent meanings (like homophones, idioms, slang, etc). Fastest way to accidentally insult someone
December 21, 2025 at 4:34 PM
Oh sorry, one more
I would not ever take the output of any software translation as accurate. Even excluding hallucinations (and that's a big exclude), what you're not getting from it is context dependent meanings (like homophones, idioms, slang, etc). Fastest way to accidentally insult someone
I would not ever take the output of any software translation as accurate. Even excluding hallucinations (and that's a big exclude), what you're not getting from it is context dependent meanings (like homophones, idioms, slang, etc). Fastest way to accidentally insult someone
ai is just guessing the most likely next word in a sentence based on its input data. it’s just smart autosuggest. there’s no logic involved to determine if the output is factually correct. the greater presence of a token in its input data means the greater likelihood it will be in the output.
December 21, 2025 at 9:01 PM
ai is just guessing the most likely next word in a sentence based on its input data. it’s just smart autosuggest. there’s no logic involved to determine if the output is factually correct. the greater presence of a token in its input data means the greater likelihood it will be in the output.
I’m pretty confident the output will improve but I’m less confident that ease of access will. this is not true for most technologies!
December 21, 2025 at 5:31 PM
I’m pretty confident the output will improve but I’m less confident that ease of access will. this is not true for most technologies!
Llm's are only useful if you don't particularly give a fuck whether your output is actually correct and it doesn't really matter if it is.
Which says a *lot* about the jobs and lives of the people who are the loudest about advocating their use.
Which says a *lot* about the jobs and lives of the people who are the loudest about advocating their use.
December 21, 2025 at 7:04 PM
Llm's are only useful if you don't particularly give a fuck whether your output is actually correct and it doesn't really matter if it is.
Which says a *lot* about the jobs and lives of the people who are the loudest about advocating their use.
Which says a *lot* about the jobs and lives of the people who are the loudest about advocating their use.
We made a mistake by calling the output of AI ‘slop’. In academia it is now looking much more like toxic pollution, on the verge of laying waste to previously fertile plains. Suing for royalties is only a small part of the battle.
December 21, 2025 at 4:49 PM
We made a mistake by calling the output of AI ‘slop’. In academia it is now looking much more like toxic pollution, on the verge of laying waste to previously fertile plains. Suing for royalties is only a small part of the battle.
😳 A newly-minted attorney used GROK to assist in drafting and research for briefs and memoranda filed in state and federal court “without verifying the accuracy of the output”
1/
cases.justia.com/federal/dist...
1/
cases.justia.com/federal/dist...
December 21, 2025 at 10:10 PM
😳 A newly-minted attorney used GROK to assist in drafting and research for briefs and memoranda filed in state and federal court “without verifying the accuracy of the output”
1/
cases.justia.com/federal/dist...
1/
cases.justia.com/federal/dist...
The courts have ruled multiple times that AI output isn't copyrightable because it has no author meaning:
1) AI writing is public domain. You don't own any of it.
2) You can't be sure that the writing in your piece isn't stolen, because authorship be damned, you are responsible for what you sell.
1) AI writing is public domain. You don't own any of it.
2) You can't be sure that the writing in your piece isn't stolen, because authorship be damned, you are responsible for what you sell.
I can't understand how "writing is my dream job" and "ugh, can't the robot do it?" coexist comfortably inside the same skull, but a lot of people live in cognitive dissonance.
December 20, 2025 at 10:15 PM
The courts have ruled multiple times that AI output isn't copyrightable because it has no author meaning:
1) AI writing is public domain. You don't own any of it.
2) You can't be sure that the writing in your piece isn't stolen, because authorship be damned, you are responsible for what you sell.
1) AI writing is public domain. You don't own any of it.
2) You can't be sure that the writing in your piece isn't stolen, because authorship be damned, you are responsible for what you sell.
‘Acceptable without explicit citation’ (of the LLM’s output). I read the entire report. You might want to yourself before insisting that this is not saying what it’s clearly saying.
December 21, 2025 at 1:51 AM
‘Acceptable without explicit citation’ (of the LLM’s output). I read the entire report. You might want to yourself before insisting that this is not saying what it’s clearly saying.
Pretty happy with my painting output this year considering I spent almost every waking moment of 2025 grinding out overtime at work.
#warhammercommunity
#paintingwarhammer
#miniaturepainting
#warhammercommunity
#paintingwarhammer
#miniaturepainting
December 20, 2025 at 4:28 PM
Pretty happy with my painting output this year considering I spent almost every waking moment of 2025 grinding out overtime at work.
#warhammercommunity
#paintingwarhammer
#miniaturepainting
#warhammercommunity
#paintingwarhammer
#miniaturepainting
THE VOICES needed an output
With all this awful news going on i wanna ask this question once again to fellow artists and general creatives. What got YOU into art? what sparked that motivation to start writing? what made you wanna pick up a camera and start taking pics? Like share with us your reasons.
December 19, 2025 at 12:18 AM
THE VOICES needed an output
Python Coding challenge - Day 917| What is the output of the following Python Code?
Answer with Explanation: www.clcoding.com/2025/12/pyth...
Answer with Explanation: www.clcoding.com/2025/12/pyth...
December 20, 2025 at 6:23 AM
Python Coding challenge - Day 917| What is the output of the following Python Code?
Answer with Explanation: www.clcoding.com/2025/12/pyth...
Answer with Explanation: www.clcoding.com/2025/12/pyth...
GasBuddy today tracking ~135 gas stations below $2 per gallon, or 0.2% of all U.S. gas stations. Cheapest price $1.81/gal in Colorado Springs, CO. Thank low oil prices and above average refinery output.
December 21, 2025 at 1:58 PM
GasBuddy today tracking ~135 gas stations below $2 per gallon, or 0.2% of all U.S. gas stations. Cheapest price $1.81/gal in Colorado Springs, CO. Thank low oil prices and above average refinery output.
The beginnings of a retro game console setup. My old Intellivision II being the first console. Waiting on an upscaler device to output to a modern display.
December 21, 2025 at 10:45 PM
The beginnings of a retro game console setup. My old Intellivision II being the first console. Waiting on an upscaler device to output to a modern display.
every artist deserves to enjoy their output, even when you don’t think your art is good, you’re probably at least one persons favorite artist
December 20, 2025 at 9:56 AM
every artist deserves to enjoy their output, even when you don’t think your art is good, you’re probably at least one persons favorite artist