Amen, and I feel exactly the same way. Using it feels awful. It feels seriously user-hostile: Its language is almost totally opaque, it takes every opportunity to criticize you and raise objections to what you give it, it virtually never gives you what you want (you have to guide it firmly and with precision) and it never substantively changes its behavior. Even its apologies, to the extent that matters, have to be forced out and are as thin as you can get when they do arrive.
It feels like it looks down on you and as if it's trying to poke holes in whatever you give it. Over time, i.e. using it for hours, it gets draining in the same way being stuck with a very clever, willfully contrary narcissist would be draining. And it's worse than with a human version -- humans eventually get tired and lose focus. But this thing can always keep churning out tokens, and using it is essentially generating irritation and psychic damage on demand.
It's surreal to be talking about a tool in these terms, but here we are. I'm switching to OpenAI myself, trying to coax a normal personality from this is absolutely not worth it.
Out of context it does sound like AInese. I think that’s largely because AI has a tendency to use location in very metaphorical terms that don’t quite read, and usually the idea of “relocating” a word would be exactly that kind of vaguely weird metaphorical use of location.
However in the context of this article “relocated the word” is both literal and precise. The word’s target is in a different physical location.
“Quietly” is also AI-coded but does fit in the overall aesthetic of the article.
As a non-native English speaker I am both astonished and terrified. Astonished that you people hear or see this difference, which is undetectable for me. Terrified that people learn English from the internet, and soon these unnatural constructions will become mainstream for us.
EDIT.
Also by using and contacting with AI won't English native speakers get accustomed to that AI-speak?
I know you're being downvoted, because people are tired of hearing "this is AI".
But I think your comment "this is AI" is warranted. This content, explaining something of cultural, personal significance (instead of the typical technical explainer) should really be written by a human.
> Possible false positive because of the text was translated from Japanese to English with AI?
People say this a lot, but I've never seen Claude or any other LLM insert clichéd LLM-isms into translated text.
There's obviously a degree of interpretation and "smoothing" that goes into translation, but it doesn't rise to the level of making the entire thing sound like an AI's voice. This blog post was prompted, not translated.
If I have read the article, then reached the comments sections, it's because I have found the subject interesting. It's like watching a movie and someone trying to tell me unsolicited news about the director's private life. I don't care: tell me about the movie, instead.
In the case of a blog that purports to be personal reflections, commenting that it is actually AI output is a comment on the content, not merely the form.
The topic is the expression "itadakimasu". What the blog purports to be, belongs in the boundary. Who wrote it belongs in the boundary. Whether the syntax is correct, contains em-dashes, uses nice colors and fonts, belong in the boundary. Comment about "itadakimasu" are on topic, others are off topic.
This is worthy feedback to the author (and others), that writing with clear AI tells/register is so distracting that it takes away from the spirit and content of the writing.
Perhaps this is just the zeitgeist and in 20 years this is irrelevant. It’s relevant for whoever is writing now.
You can say the same about detecting grammatical mistakes: it's nice to warn the author, but please send them a private message, don't pollute the comments section.
Pangram's actual classifier model is really strong. It has an incredibly low false positive rate. Whenever people complain about it, I ask them to provide verifiable non-adversarial long-form false positives, and nobody has done so yet. False negatives are more common (as they should be -- it's better to let an AI article through than to false accuse someone.) Their frontend is often misleading, though, for these reasons:
- 100% means that it thinks 100% of the article is AI, not that it's 100% confident the article is AI. It's very easy for people to get this confused.
- It is much less accurate on shorter text, and doesn't advertise this fact clearly enough.
- It works on chunks of text (I think they're like 100 tokens) and can't identify within that chunk which parts are AI and which parts are human.) This should be more clear in the UI, I think.
This is good, but I think it's missing a discussion of the business logic, which is IMO the mechanism behind the gravity we're discussing here. And really it's just about the fact that you always get more positive attention in an org (and therefore a better shot at promotion etc.) for suggesting and implementing new stuff than pruning what's already there. Pruning only becomes possible when there's a coherent vision, for which you need a small-to-medium team (at most) and one person with the vision in charge of it.
Wft. I have a piddly little Claude Pro account, I use it every day, and Opus 5 and Fable (judiciously used) are very helpful and useful in my work and personal life. I am happy with my setup. Then I read this, and I imagine it's what primitive farming societies felt like when introduced to combine harvesters, high-clearance sprayers and mega tractors.
I am happy on my farm. I like the carrots I plant by hand. I do not wish to control farming equipment worth a small fortune to till acres of land every day.
(P.S. Still not sure if TFA is satire! Obviously, this assumes that it's not.)
No, the issue was that he apparently took direct quotes from participants in another study, reworded them a little, and presented them as quotes from his own participants. This is all in TFA.
Very good too, actually, assuming you enjoy the premise of CYOA in the corporate world. You can play it online for free if you want to give it a whirl.
Well he just « overhauled » them… which I’m going to interpret as laying off a bunch who didn’t flag this to him earlier. If you just got hired by a billionaire who fired everyone on the prior team for not catching something, and is paying you to be brutally accurate are you going to order 1 biopsy? 2?
One gives you a high potential for false positives or negatives, two data points with opposite results is just as useless. Three is the gold standard to create a consensus among sensors, a lesson Boeing learned with their angle of attack at the cost of many lives.
Not sure, I would say it overstates things anyway (it's Twitter, so exaggeration for engagement is par for the course) ... but I would say it's gesturing at something roughly true and it'd be worth discussing the idea.
> Our elbows graze and his windbreaker says “vwip,” “vwop,” the walk feeling almost like a jog until finally he slows, we’re getting ready to jaywalk
The little onomatopoeia feels quite forced and out of place here -- totally the wrong note. FWIW, AI really likes writing that kind of thing, and I wonder to what degree that's influencing actual writers.
> Marshall McLuhan gets the credit for the medium is the message, but Claude Shannon had beaten him to a colder version of it years earlier: to a machine moving your words, the meaning doesn’t matter at all; only the medium does, and which of its signals can be told apart. Bravo and Delta survive a bad line; B and D don’t.
> I didn’t arrive there as a mathematician; I’m not one.
> This wasn’t a speed problem I could optimise away. It was a wall, and it asked a question I couldn’t answer
Very strong LLM whiff. A line of thought that constantly, constantly turns back on itself, negating and doubting and qualifying in one way or another, is the biggest tell (the classic "It's not X, it's Y," is only the baldest example).
Noticing that whiff instantly turns me off from reading on.
A few hours ago, on another post, I had someone explain/teach me why another article may be written by LLM and they pointed out the "It's not X, it's Y" thing. Since I read that comment, I instantly picked that up in this post article too.
which really sucks, because it's a good structure for emphasis in some limited situations, but now human writers that use it correctly get accused of being LLMs
> Yes but I'm fairly sure you didn't use the words "this isn't a speed problem. It is a wall" when talking to your partner.
I am not afraid of a future where people use llms to write. I am afraid of a future where people adopt themselves the writing style of llms because that's all they ingest.
Yeah that's fair. I might dust off my draft and re-edit myself, I foolishly thought this would lead to a better post, and I guess am not as attuned to the AI smell as others.
Yeah I'm now in the camp where it's just refreshing to read something another human wrote, even if it has mistakes, just because I'm tired of only reading a single author everywhere.
I’ve noticed that A.I. can’t help but talk about Marshall McLuhan when you mention “medium” and “message” for some reason. Personally I don’t find the guy’s “medium is the message” bit to be so insightful as to merit pinning his name to every mention of the phrase, forever. It just seems kind of obvious.
Sorry about the turn off ¯\_(ツ)_/¯ I tried to put my best foot forward by reading about prose, engaging story telling, and did use an LLM to help me edit and reword parts of the post. Either way, I appreciate the feedback.
It feels like it looks down on you and as if it's trying to poke holes in whatever you give it. Over time, i.e. using it for hours, it gets draining in the same way being stuck with a very clever, willfully contrary narcissist would be draining. And it's worse than with a human version -- humans eventually get tired and lose focus. But this thing can always keep churning out tokens, and using it is essentially generating irritation and psychic damage on demand.
It's surreal to be talking about a tool in these terms, but here we are. I'm switching to OpenAI myself, trying to coax a normal personality from this is absolutely not worth it.
reply