I love how this waxes poetically how "The exact words we choose when writing matter.
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante @daringfireball oh dog, it’s sooo looooong, I almost caved and got an AI to summarise it for me!
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante @daringfireball The problem with dismissing this as “nanny state, pipe-dream nonsense”, like Gruber does with basically everything coming from the EU these days, is that he fails to engage with *why* the EU is doing this. If the EU wants it to be reasonably obvious that any content was AI-generated, it’s because they think there otherwise would be too much AI. Ergo, making AI-generated text worse is exactly what the policy demands!
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante
Reading texts is part of my profession.As I tried to read that extreme tl;dr I asked myself, if my brain had lost too many synapses by the heatwave.
If the future are such texts, 🥶 I want to retire straight away!
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante I'm particularly bewildered at the part about proofreading: so how do people use this stuff for proofreading exactly? Does Gruber accept el Claudillo's decree's^Wedits willy-nilly? "Oh the parrot likes this word better, I guess I'll use it then". Well, then you do deserve being flagged as AI-generated!
-
@tante "The idea that anything other than *my* needs should factor into the generation of text for *me* is patently offensive."
That's the most direct "I don't care about externalities" I've read in a while.
-
@tante
Reading texts is part of my profession.As I tried to read that extreme tl;dr I asked myself, if my brain had lost too many synapses by the heatwave.
If the future are such texts, 🥶 I want to retire straight away!
-
@tante @daringfireball oh dog, it’s sooo looooong, I almost caved and got an AI to summarise it for me!
-
-
His brain is truly cooked
"LLMs are, in their popular incarnations, non-deterministic. Ask the same question of the same model and you often get at least slightly different answers. Maybe the same meaning, but different phrasing"
"One of my fundamental problems with [watermarking] is that no two synonyms carry the exact same meaning. “He leaped at the chance” and “He jumped at the opportunity” are very similar sentences expressing the same general sentiment, but they are not the same
@frog_reborn @tante lmaooo. words only mean things sometimes, when it's convenient for my argument
-
@tymwol @danielaKay @tante Was going to say this also — patterns are likely to transfer from llm to natural writing as people get exposed to the former.
@GerardThornley @tymwol @danielaKay @tante
The watermarking is about *how* you choose the next token, not *which* word that ends up being. So the patterns are not something you can recognize or subconsciously learn and repeat.
For a gross oversimplification, imagine you always choose the second most likely candidate exactly eight times in a row, followed by selecting the fourth candidate exactly once, followed by exactly five more second candidates.
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante Yes, the genocide fan who denies trans voices and looks down on most of humanity is also a botlicker.
Anyone non-critically linking to independent Fox News dipshit John Gruber is dangerous.
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
Yes, silly to claim outrage about a non-deterministic thing being non-deterministic in a very slightly different way. It's a bit like giving an output in millimetres to a sum where one of the inputs was "about a mile".
But I feel I must give the writer credit for the phrase "the saccharine Claude personality stink"!
-
@tante I'm particularly bewildered at the part about proofreading: so how do people use this stuff for proofreading exactly? Does Gruber accept el Claudillo's decree's^Wedits willy-nilly? "Oh the parrot likes this word better, I guess I'll use it then". Well, then you do deserve being flagged as AI-generated!
-
@tante I'm particularly bewildered at the part about proofreading: so how do people use this stuff for proofreading exactly? Does Gruber accept el Claudillo's decree's^Wedits willy-nilly? "Oh the parrot likes this word better, I guess I'll use it then". Well, then you do deserve being flagged as AI-generated!
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante Also, you want the "best" word? What an unsolvable philosophical predicament to put onto a computer. Saying such things makes it obvious that they do not care about writing, language, or intent.
-
@tante
To me it's also like, not everything has to be perfectly optimised. How you write *doesn't* have to be perfect. How you live life doesn't have to be optimised. -
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante a perversion of writing? By the perversion of writing machine?
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante I totally identify with the sentiment. When I write, I want to result to reflect my intended meaning as faithfully as possible. I just don't get how you ever expect that from a LLM, even before it starts using acrostic poetry to thread its own copyright through what is ostensibly your writing.
-
@danielaKay @tante you cannot watermark spelling, so it you used it only for spelling, it's a non issue...
To be honest, I'd rather be worried it'd be the other way around: they watermark it by using specific words, etc, so it is a bit like people were treating em-dashes as a sign of AI usage, just now it is done on purpose. So same as before you could get accused of using AI because you used an em-dash, now it could be something else. So yeah, it would likely lead to some new problems we never needed.
In general, there is a massive amount of harm being done by normal numbnuts going "this is clearly AI, look at the [random shit]".
Because most of the "telltales" of AI are due not to some "watermark" but to theft.
The AI company stole some people's work, and now those people are being accused of using "AI".
There are at least two artists on here having to deal with that on the regular, and it must be so infuriating and awful and disspiriting.
-
RE: https://mastodon.social/@daringfireball/117106850045916140
I love how this waxes poetically how "The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point."
My brother in christ: Then you shouldn't be using a stochastic text extruder.
@tante
I would submit that if you consider the mythical average person and their writing, that an LLM would almost certainly improve that writing
