"I did not like your answer, therefore I will use the 100% reliable, bullet-proof method of having an algorithm generate the statistically most likely words that form a plausible answer to my question."
Except that the entire point of the article is that they're not AI idioms. They're not "watermarks for text." They're legitimate language constructions that LLMs tend to overuse, but that real humans also use. Real humans do, in fact, say "align with" all the time, just as often as "corresponds."
And you can pry my em dashes from my cold, dead hands.
What's worse is neurodivergent writing, including my own, often resemble AI output. Now it feels like I'm having to alter my own voice in online discussions just to specifically avoid being accused of pasting an AI response.
The "AI Detection" tools employed by schools also regularly flag writing from those with Autism, ADHD, and non-native English speakers as being AI generated as well.
So, naturally, I can't stand the phrase "write like AI" when these things tend to come up because no, there are no humans that "write like AI" it's the models that have stolen the literary devices from us and now have poisoned them.
> While the detectors were “near-perfect” in evaluating essays written by U.S.-born eighth-graders, they classified more than half of TOEFL essays (61.22%) written by non-native English students as AI-generated (TOEFL is an acronym for the Test of English as a Foreign Language).
> It gets worse. According to the study, all seven AI detectors unanimously identified 18 of the 91 TOEFL student essays (19%) as AI-generated and a remarkable 89 of the 91 TOEFL essays (97%) were flagged by at least one of the detectors.
The article is not God, just because it claims something doesn't mean we have to accept it.
For better or worse (and pretty much for worse), these usages have become AI idioms. Language evolves over time, things that used to be harmless become offensive, certain terms end up taking on the complete opposite meaning than their original meaning, and we are watching certain language patterns and idioms become watermarks for AI and while it sucks, it doesn't make it false.
I'll just quote from the article, which no one claimed was God and that's really a weird way to dismiss it, but you do you:
"We create a culture of self-censorship and AI-detector-pressured rewriting and paraphrasing as people strive to avoid these witch hunts. That is the opposite of protecting human expression. We should resist normalizing a trust in any machine's ability to determine matters of guilt. If using AI to write is, at its worst, an industrialization of the mind, then AI detection, at its worst, becomes a surveillance system for thought."
And, I'm sorry (I'm not), but I am not going to just roll over and shrug and say "welp, guess we all need to dumb our writing down to keep well-meaning idiots from screeching 'AI! AI! AI! WHOOP! WHOOP WHOOP WHOOP!' at us." That isn't the evolution of language. It's Idiocracy.
Well reading between the lines I don’t think they’re saying all of those uses are AI. They’re legitimate constructs, like the em-dash, en-dash, and hyphen, all of which I used to use regularly. But now they’re AI tells so I use them sparingly.
that's the thing. em-dashes are such a low hanging fruit, that i fear that they soon will be gone. i mean if i am ever to post an AI generated text, removing them is the first thing i would do.
(edit: turns out it doesn't help as much as i thought it would. see the edit above)
I mean, a company that loses money on every widget they sell might technically have found "product-market fit." :)
It seems quite possible to me that developer tooling is going to end up being the biggest win from LLMs because there is a product-market fit -- and also quite possible that OpenAI and/or Anthropic end up getting bought for pennies on the dollar because their burn rate is unsustainable. AI may end up being this generation's "dark fiber."
Congratulations to last.fm on this, although I'm not sure what I can do that actually "scrobbles" at this point. I went to their download page for the Mac desktop app, which somewhat forebodingly referred to "what you've been playing in iTunes" rather than Apple Music, and downloaded it anyway. It's an Intel-only app, not universal, and it doesn't appear to be signed, so macOS Tahoe screams about it being possible malware.
After going through the hula dance to open it anyway, it looks like it's working, but it sure doesn't look like it's received a lot of love recently.
Spotify scrobbles to Last.FM (it's built-in to Spotify as a setting somewhere, from memory). On Windows, MusicBee will scrobble everything you play through it. On Android, the Last.FM app can watch playback notifications of Android music apps that you select and scrobble those - the app itself doesn't have to have Last.FM support.
I've tried to switch to Spotify from Apple Music a few times because the common wisdom seems to be that Spotify has better algorithmic recommendations. But Apple Music "knows" what I like already, and Spotify never grabs me so fast that I'm willing to stick around for weeks training it -- and I suspect part of that is all of Apple Music's human-made playlists. Apple Music has hired a lot of good editors/curators over the years, and I haven't found any service -- including audiophile darlings Qobuz and Tidal -- that beats it in that aspect.
There seems to be a "spot the LLM" game happening on HN in particular in which literally every damn linked post has a comment accusing it of being written by an AI. Do we not understand that definitionally every AI "tell" comes from humans? Hell, Sam Kriss's article complains about florid writing and then goes on to cite Salman Rushdie and Arundhati Roy as examples of writers who "pull the same cheap tricks" as AI and with all respect to Mr. Kriss I just can't even. The theme of this article veers dangerously close to "if you use a weird metaphor you're probably using an LLM because surely no human would ever write 'I'm literally the size of a dry martini.'" I'm sorry, Sam, but humans have written weird fucking shit for a very long time without LLMs.
And what "tells" are in this article, anyway? It reads very straightforwardly; it does not, in fact, have any weird metaphors. Em dashes are not actually a tell, but using the wrong characters for em dashes (the article uses " - ", e.g., space hyphen space) seems pretty human. Last but, I would argue not least, I am willing to give the author of an article about not wanting to program with LLMs the benefit of the doubt here. "The orphan-punching machine fills me with existential angst and dread, and the only way for me to communicate this to the audience is to let it punch a few orphans for me."
Now, if you'll excuse me, all this is making me need to be literally the size of a high-proof daiquiri.
I'm not sure. It might be a real person, just writing in an LLM-like hackneyed voice. Anyway here's three specific paragraphs from the post, each exhibiting that voice and the "rule of three":
> Throughout all of these experiences, and so many more than I have time to share, it was the human connection that made it special. The laughter that helped me get through a hard problem. The sleepless nights that reminded me I was not alone. The selflessness of others to get on stage or behind a camera and teach people they didn’t even know, often for little or no cost, just so that others would be enabled to build and have an influence.
> They don’t laugh with me when my code fails to compile after I swear “this is the one.” They don’t help me develop an understanding of my software, so that when someone says “how does this work” I can pour my heart out with passionate explanations. Most importantly, they don’t turn their head and smile and participate in the inexplicable elation of saying “we built this!”
> I desire to connect with people. I long for the days where I was vulnerable and shared my struggles with engineers who charitably stepped up to support me. I miss taking what I learned from those struggles and sharing them back out as a blog post or presentation, encouraging the next person to overcome the same challenge.
This seems to be willfully eliding that proposed wealth taxes tend to either be taxes on wealth above a certain amount, or (such as California’s) a one-time tax on people with wealth over a certain amount. If I were a mere millionaire -- technically, I am, with a net worth of just over $1.1M, but this would be true if that were $5M or $10M or even $50M -- then under any proposal I’ve seen, my wealth tax would be $0. (Note that if someone were to have $50M, then under Graham’s risk-free rate of return of 5%, they would literally have to do nothing to pay themselves an “income” of $2.5M annually.)
If I were an actual billionaire -- say, my net worth was $2B -- then my one-time tax under California’s proposal would be $100M, leaving me with a net worth of $1.9B. Under that 5% risk-free rate of return, I would recover that amount of money within one year even if my income were $0, which seems exceedingly unlikely.
One can argue about the specifics of various proposals -- the Tax Foundation, for example, thinks California’s proposal has “aggressive design choices and possible drafting errors” that could lead to somewhat bonkers results, although I haven’t seen any critiques of their analysis yet -- but a wealth tax cannot be converted to income tax in a reasonable manner any more than a VAT could be converted to property tax. They’re both taxes, but they’re simply not the same kind of tax. And while I don’t mean to cap on Paul here, there’s a distinct “woe, pity the poor billionaires who will surely be driven to bankruptcy” subtext I find to be risible nonsense.
The traditional name for a surprise "one time wealth tax" is a capital levy. It's got a pretty terrible reputation all around because it's the closest thing to an official declaration that your country (or state as the case may be) is now a complete fiscal and financial basket case that can't manage to fund itself by sensible means.
No one believes or acts like this will be a one time event (on any side of the issue). The history of all new forms of taxation is that eventually it will come for you.
I admit I have recently moved, on my Mac, from BBEdit to Emacs for most things, although Emacs has also taken over from Ulysses and Day One for me. But there are still some things that I end up doing in BBEdit, because it's kind of a multitool for text manipulation that, unlike Emacs, exposes its power very easily. As much as I'm growing to love Emacs, there's something to be said for an editor that doesn't answer so many questions with "write a little Lisp".
It seems to me (entirely anecdotal, YMMV, etc. etc.) that Ed Zitron’s blog posts started getting both longer and considerably more histrionic when he started moving most of them behind a paywall. I’m definitely in the “AI skeptic” camp and think Zitron has good points to make about both the shaky business models around AI and the unrelenting hype train, but it’s hard not to get the impression that he’s found a niche of preaching to the rabid AI haters willing to give him money to keep spouting increasingly repetitive vitriol toward Sam Altman and Dario Amodei.
reply