Rendered at 10:51:24 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
karmelapple 10 hours ago [-]
So glad to see color photographs of the site.
One thing I'm not so glad to see is one paragraph that starts with this sentence:
"The distinction is important. Archaeological maps..."
That "The distinction is important" smells like AI writing... but maybe we all just have to accept that now?
azernik 5 hours ago [-]
We all have to accept that people will look at perfectly innocuous, clear, well-written prose and assume that AI wrote it.
fennecfoxy 14 minutes ago [-]
For sure. If you write with a casual conversational tone people don't get triggered. But if you are eloquent and use an extended vocabulary thanks to large amounts of reading then that seems to make people think you're using AI.
As someone who loves to read (well, on and off) I wonder what the median vocabulary is and how I and other readers compare. I guess language has always been functional for most people.
kakacik 3 hours ago [-]
Yeah its getting annoying how people discuss endlessly how some text is AI or not... like some extra imaginative points are awarded to those who spot the most. People seem to keep getting triggered by little insignificant things in the text.
LLMs average human writing thats out there, its not making up new expressions or language constructs so just move on if you (anybody) can and put that little obsessive pedantic part of you behind you.
I couldn't care less, why be a prick and not just enjoy & discuss the content, not this ultra focus on the form. If content is crap then thats another story altogether. I wish there would be some plugin/feature to just block those rather empty subthreads.
chmod775 2 hours ago [-]
> LLMs average human writing thats out there
Not really, and also they do not vary their writing like humans would. Human writers do not repeat the same pattern over and over as much as LLMs do. They'd get bored of their own writing.
People as a whole are always going to be sick of whichever affectation current popular LLMs have, since they'd be reading a fair bit of that. As a writer you should steer clear of things that would elicit groans in a contemporary audience and "LLM prose" is just the latest addition.
So, as the person you're replying to said, you'll have to get used to this being a new angle from which writing is critiqued. You'll live happier if you don't let your cortisol spike over it. If it is any comfort, writers will eventually adjust and you'll encounter less of it.
fennecfoxy 8 minutes ago [-]
Interesting. I agree that it seems that LLMs are wont to fall into recognisable patterns much more readily than humans however I asked a model to evaluate your comment history and it found loads of patterns such as:
“I’m not sure…” / “I don’t…” hedges before disagreement
Negation followed by correction/reframing is a recurring structure: essentially “It’s not X, it’s Y”.
And dozens of other examples. I wonder if LLMs really are more liable for it or if it simply seems that way purely because when you use a model it's almost as if you're speaking to the same "person". I think that human language is naturally formed by common agreed meanings of words first, and then phrases.
3 hours ago [-]
psychoslave 4 hours ago [-]
One can still suspect that something might not have been uttered through some AI assistance. There might even be still some living humans outside, who knows.
0xpgm 10 hours ago [-]
As more people start using AI for research, they'll probably adopt AI'isms in their writing too, even if they don't use AI itself to do the writing.
Subconsciously we all pick up patterns that we regularly encounter.
hunter-gatherer 9 hours ago [-]
I've had that happen to me. I actually recently wrote a draft blog and looked over it, and it felt AI generated. My writing style has definitely changed.
seer 3 hours ago [-]
Hah absolutely, after of a very intense project I’ve undertaken with Claude agent teams for a few months now, I have caught myself using opus expressions, since they come to mind now naturally because that’s most of the text I’ve read since then … I’m catching myself so far, but in a year or two maybe I will not, fable doesn’t do as much “mannerisms” and speaks plainly, so maybe by then we will all be running fable level models on hour local machines and just not notice it anymore?
jcims 7 hours ago [-]
I've definitely picked up 'lever', I scrub it every time that I notice I write it but it's in there.
rubicon33 4 hours ago [-]
“blast radius” for me
youniverse 8 hours ago [-]
Maybe we need our AI to only respond in bullet point lists.
erickhill 7 hours ago [-]
With emoji for the bullets, natch.
TomK32 6 hours ago [-]
Just put "Don't use emojis" into the prompt.
tacomagick 4 hours ago [-]
Pass it as a memory or a system prompt and watch it forget 3 agent messages later.
mxmilkiib 4 hours ago [-]
what frontier models are this bad these days?
tacomagick 3 hours ago [-]
I can't name one but I can tell you these models keep getting stupider as they keep existing and their system prompts get inflated with what not to do. I have experienced this happen with Qwen 3.8 Max Preview where for the early days it was blasting through Rust code fine and then before the release it silently got replaced and started to suck hard. Claude still has the "blue button" problem so it is probable that a casual chat with it could have this exact outcome of forgetting what to do. I saw some models also have this thing where when you tell them a text restriction they first ignore it in thinking, creating the full text and then apply them in their thinking rewriting it fully causing your output tokens to be inflated.
fennecfoxy 2 minutes ago [-]
Well yes, there are only so many attention heads (well whatever magical variant frontier models are using these days) that can attend to the context and so as the context grows attention becomes spread thin.
But with reasoning enabled I find that even with a large context that induces mistakes most of the time a decent model realises and corrects itself before output.
And it's always been known that prompting what it should do is far better than what it shouldn't, since just introducing "DON'T do X" into the prompt means that the tokens for X are present and can be paid attention to in the wrong way.
But even then I've used plenty of "Do X, not Y" recently, especially for tools "Use this for x, don't use this for Y" and models perform like 90% of the time.
I would be interesting to experiment to see performance curves given a restriction on reasoning tokens allowed to n% of context tokens and see if there's some magic number of "reasoning should be at least n tokens for a context of length p" even ignoring the complexity of instructions in the prompt itself.
But I still find that I have to consciously fight the urge to sound like an LLM. It's terrible.
joegibbs 6 hours ago [-]
I have the opposite, I can pick up AIsms much more now and they stick out like a sore thumb. A couple of years ago I had no such intuition, it just sounded like normal writing to me, but now if I try to use AI to write any prose it just sounds absolutely terrible.
awesome_dude 5 hours ago [-]
This is the load-bearing result
Razengan 3 hours ago [-]
You're right to call it out.
debo_ 9 hours ago [-]
This distinction is important.
21asdffdsa12 4 hours ago [-]
Remember the "IDunno" of archeology - "It most likely served ritual and cultural purposes" .. i wonder if AI adapts such a thing
st_goliath 3 hours ago [-]
> the "IDunno" of archeology - "It most likely served ritual and cultural purposes"
If all you know about archeology is from daytime television and clickbait articles, all you know about history is what you vaguely remember from school, you never met anybody working in those fields and actually don't even have a passing interest to engage with the subject, then yes, you might be inclined to believe that's how it works.
EDIT: of course this comment accumulates nothing but drive-by down votes from arrogant computer touchers, presumably before heading off to other comment sections to discuss law, macroeconomics, quantum physics and whatever else they are experts at, having watched YouTube videos. You see, only CS requires years of study and is divided into dozens of intricate sub fields, everything else is simple, straight forward and can be "disrupted".
jimmygrapes 3 hours ago [-]
Worst thing is that because the smell is bad and you called it out, too comments are all about the smell vs the substance. I contribute, sadly. Maybe we should all be more judicious with when we say there's a smell, save that for when it's really rank? Like a polite thing. Yeah it stinks but maybe it's worth something beyond that. Not sure if this is the one, just thinking out loud.
TomK32 6 hours ago [-]
Ran it through several testes, only one flagged a few sentences as AI, including the one you mention. Fair enough, it's unlikely that English is the (turkish in this case?) author's first language and a little AI help can't hurt here.
I learned a great lesson for a gamedev youtuber this summer: Use AI to give you a summary what you have to implement, what functions you'll need, what edge case you missed and then write it yourself. Putting this into practice helps me to refine the ticket I'm working on. With the last ticket I had to persuade the AI (Atlassian's Rovo in my case) that a new data model wasn't necessary and we can do it more elegantly with less code.
Many of us still have to figure out how AI fits into our workflow, letting it do all the work isn't it but asking AI to point you towards the goal and let it review/refine your work sounds a lot more acceptable to me.
Bad writing makes the reader stop thinking about what you wanted to convey. As unintuitive as it is for most people in this forum, AI writing smell is as bad as bad grammar. Doesn't matter how you get rid of it. Maybe even more AI can clean those up, but those unintuitive phrases that pop up in AI writing are going to be a hindrance for many people who would love for the focus to be on what they're presenting, not the sloppy way they've written it up
pitkali 3 hours ago [-]
> AI writing smell is as bad as bad grammar
It sounds to me like you want to censor perfectly fine phrasing just because AI rolled the dice to use it. Let's not forget that AI learned to "speak" from people in the first place.
Not to mention that as people shift their phrasing to avoid being accused of using AI, the AI will keep evolving, training on the new phrasing and then adopting it, too.
I understand the language evolves, but evolving language just to sound different from AI feels like wasting effort better spent elsewhere. On actual substance of the message for example.
protocolture 5 hours ago [-]
>testes
teehee
6 hours ago [-]
sho_hn 3 hours ago [-]
I'm giving a talk this Sunday at KDE's Akademy conference of our use of KDE software in research at the Great Pyramid, if anyone's curious. It's a short talk slot and I cannot go very deeply into detail, to keep expectations low, but you can see cool photos of robots.
gloflo 3 hours ago [-]
Will there be a video recording or slides?
sho_hn 2 hours ago [-]
Definitely!
tym0 2 hours ago [-]
How are things like that, that were preserved (I'm assuming) by the dry sand they were buried under, then gets preserved by archaeologist for the future?
A lot of pigments react to oxygen, and the sandy winds must not be kind to this kind stuff so I'm curious.
I know they've had to close Lascaux to the public for example.
NegativeLatency 5 hours ago [-]
I imagine they've done some tweaking before publishing (also appears the rock substrate has flaked off), was hoping to be able to pull some more detail out of the images with a vibe coded decorrelation stretch app I made from a post last week:
One thing I'm not so glad to see is one paragraph that starts with this sentence:
"The distinction is important. Archaeological maps..."
That "The distinction is important" smells like AI writing... but maybe we all just have to accept that now?
As someone who loves to read (well, on and off) I wonder what the median vocabulary is and how I and other readers compare. I guess language has always been functional for most people.
LLMs average human writing thats out there, its not making up new expressions or language constructs so just move on if you (anybody) can and put that little obsessive pedantic part of you behind you.
I couldn't care less, why be a prick and not just enjoy & discuss the content, not this ultra focus on the form. If content is crap then thats another story altogether. I wish there would be some plugin/feature to just block those rather empty subthreads.
Not really, and also they do not vary their writing like humans would. Human writers do not repeat the same pattern over and over as much as LLMs do. They'd get bored of their own writing.
People as a whole are always going to be sick of whichever affectation current popular LLMs have, since they'd be reading a fair bit of that. As a writer you should steer clear of things that would elicit groans in a contemporary audience and "LLM prose" is just the latest addition.
So, as the person you're replying to said, you'll have to get used to this being a new angle from which writing is critiqued. You'll live happier if you don't let your cortisol spike over it. If it is any comfort, writers will eventually adjust and you'll encounter less of it.
“I’m not sure…” / “I don’t…” hedges before disagreement Negation followed by correction/reframing is a recurring structure: essentially “It’s not X, it’s Y”.
And dozens of other examples. I wonder if LLMs really are more liable for it or if it simply seems that way purely because when you use a model it's almost as if you're speaking to the same "person". I think that human language is naturally formed by common agreed meanings of words first, and then phrases.
Subconsciously we all pick up patterns that we regularly encounter.
But with reasoning enabled I find that even with a large context that induces mistakes most of the time a decent model realises and corrects itself before output.
And it's always been known that prompting what it should do is far better than what it shouldn't, since just introducing "DON'T do X" into the prompt means that the tokens for X are present and can be paid attention to in the wrong way.
But even then I've used plenty of "Do X, not Y" recently, especially for tools "Use this for x, don't use this for Y" and models perform like 90% of the time.
I would be interesting to experiment to see performance curves given a restriction on reasoning tokens allowed to n% of context tokens and see if there's some magic number of "reasoning should be at least n tokens for a context of length p" even ignoring the complexity of instructions in the prompt itself.
- Distinction
- Is
- Important
But I still find that I have to consciously fight the urge to sound like an LLM. It's terrible.
If all you know about archeology is from daytime television and clickbait articles, all you know about history is what you vaguely remember from school, you never met anybody working in those fields and actually don't even have a passing interest to engage with the subject, then yes, you might be inclined to believe that's how it works.
EDIT: of course this comment accumulates nothing but drive-by down votes from arrogant computer touchers, presumably before heading off to other comment sections to discuss law, macroeconomics, quantum physics and whatever else they are experts at, having watched YouTube videos. You see, only CS requires years of study and is divided into dozens of intricate sub fields, everything else is simple, straight forward and can be "disrupted".
I learned a great lesson for a gamedev youtuber this summer: Use AI to give you a summary what you have to implement, what functions you'll need, what edge case you missed and then write it yourself. Putting this into practice helps me to refine the ticket I'm working on. With the last ticket I had to persuade the AI (Atlassian's Rovo in my case) that a new data model wasn't necessary and we can do it more elegantly with less code.
Many of us still have to figure out how AI fits into our workflow, letting it do all the work isn't it but asking AI to point you towards the goal and let it review/refine your work sounds a lot more acceptable to me.
I am not affiliated with pangram.
Wrong wrong wrong.
Bad writing makes the reader stop thinking about what you wanted to convey. As unintuitive as it is for most people in this forum, AI writing smell is as bad as bad grammar. Doesn't matter how you get rid of it. Maybe even more AI can clean those up, but those unintuitive phrases that pop up in AI writing are going to be a hindrance for many people who would love for the focus to be on what they're presenting, not the sloppy way they've written it up
It sounds to me like you want to censor perfectly fine phrasing just because AI rolled the dice to use it. Let's not forget that AI learned to "speak" from people in the first place.
Not to mention that as people shift their phrasing to avoid being accused of using AI, the AI will keep evolving, training on the new phrasing and then adopting it, too.
I understand the language evolves, but evolving language just to sound different from AI feels like wasting effort better spent elsewhere. On actual substance of the message for example.
teehee
A lot of pigments react to oxygen, and the sandy winds must not be kind to this kind stuff so I'm curious.
I know they've had to close Lascaux to the public for example.
https://news.ycombinator.com/item?id=49645437
https://github.com/nburns/DecorrelationStretch
https://en.wikipedia.org/wiki/African_humid_period
I guess we'll find out when the next cycle blows the sand away?