[simonw]

"There is no independent audit, no time series, no disclosed methodology, so we have no idea whether the real figure is higher, whether it is growing, or how it compares across the other frontier models, none of which publish equivalent data."

Tip for writers: aggressively filter out the "no X, no Y, no Z" pattern from your writing. Whether or not you used AI to help you write it's such a red flag now that you should be actively avoiding it in anything you publish.

[falcor84]

Why is it a red flag?

How is it different from any other purely stylistic rules such as Strunk and White's prohibitions against split infinitives and the passive voice, which we've left far behind us? Why shouldn't people just write however feels natural to them as long as the message is clear?

[simonw]

Because LLMs use it constantly, to the point that it sets my teeth on edge and instantly makes me question if reading the piece is worth my time.

[falcor84]

But LLMs were literally evolved via RLHF to write in a way that humans find agreeable. Can't we just move past this aversion and accept "writing like an LLM" as generally good writing style advice?

[simonw]

The reason this particular quirk annoys me so much is that it isn't good writing advice.

Consider the two examples from this article (which may well have been human-written for all I know):

"These numbers come from OpenAI itself. There is no independent audit, no time series, no disclosed methodology, so we have no idea..."

No time series? That's non-sensical to me, it feels like that's there just to fill the quota of three things. Plus why would we assume an "independent audit" until told otherwise?

Then in the weird table, for "Institutional infrastructure" against "Personal AI safety":

"Scattered across psychology, HCI, education, and clinical informatics departments. No dedicated institute, no named fellowship, no equivalent job board."

Again, "no X" in a pattern or 3. And non-sensical - why would the fellowship be named?

It's word salad, there to fill a three-nos quota.

[falcor84]

Yeah, no, I absolutely agree with you that TFA is not an examplar of good writing. But would just argue that the problem has little to do with these snowclone patterns or the rule of 3, and a lot more with the actual substance not fitting the form, and arguably not being substantive at all.

I'm all for rejecting bad writing and bad reasoning, but just wouldn't us as a community to get into the habit of rejecting otherwise good writing just because it's AI-ish.

[amoe_]

The difference is that these rhetorical techniques need to be used with taste. LLMs just sprinkle them everywhere to try to make their copy sound good, even when it's completely inappropriate tone-wise. They don't make higher level judgements about when to employ specific features.

[Yokohiii]

Idk, I remember that writing pattern from GPT, but not from Gemini.

[mitjam]

… and “That’s not x. That’s y.” Certain LLMs wield powerful stylistic devices all the time to a point where they become irrelevant and cringe.

I see it as a good sign that we can learn to recognize the pattern and adapt but there are probably more subtle things we don’t see.

[mitjam]

I have run the piece through an impromptu stylistic device detector. It found 15 different, each used multiple times and likened the writing style as a mix of Ezra Klein, Hannah Arendt, Zeynep Tufekci, George Orwell (“especially in the contrastive clarity”).

A) I certainly don’t see enough of the tells.

B) what happens to our language if everything is written as if it’s competing for a Pulitzer’s Price?