last topazReported by @keen shore
#ChatGPT is imposing woke-style language on me
elfin panther
That's a non sequitur
conclusion does not follow from the premise
this is more of a localization issue
nova frigate
y i k e s.
keen shore
Thank you for your replies — each enlightening in its own way — which, if anything, only reinforce the need for clarification.
As outlined in my original bug report — written using a reproducible method, grounded in established French language norms and guided by a concern for linguistic neutrality — I documented a problematic behavior: the appearance of an ideologically marked form (investi·e), unsolicited, in a French-language response. This wasn’t a slip or a matter of style; it was a breach of the grammatical, orthographic, and pragmatic standards expected of a language assistant.
The first reply attempts to dismiss the report by calling it a non sequitur — as if pointing out a grammatically invalid and ideologically coded form in an AI response somehow lacked logical coherence.
Perhaps the implication is that I’ve misunderstood what “grammatical precision” and “neutral assistance” are supposed to mean these days.
It then shifts to a supposed localization issue. That explanation would only hold if investi·e were a recognized form somewhere — which it isn’t. Not in France, not in Canada, not in Belgium, not in Luxembourg, not anywhere, really, outside a few narrowly activist contexts. Unless OpenAI’s language models are now being localized for activist forums, the argument seems rather thin.
The second reply — “yikes.” — offers, in its own way, a powerful counterpoint: concise, evocative... and entirely devoid of reasoning. It’s the digital equivalent of an arched eyebrow followed by an exit — which, admittedly, is one way to try to delegitimize a valid objection while carefully avoiding any need to respond to it.
My request remains simple: a language model should default to recognized and standard usage, unless explicitly instructed otherwise by the user. Introducing fringe orthographic conventions by default isn’t inclusion — it’s substitution. It’s a political choice.
elfin panther
"ChatGPT is imposing woke-style language on me"
a) LLMs are purely statically models, they don't have intent and therefore can't "impose" anything onto you. Their creators, however, could
b) your tone and word selection is intentionally antagonistic and aren't constructive. You took this as an opportunity to voice your own opinion rather than using this as a space to discuss potential issues with the LLM
most who use the term "woke" rarely understand the full historical and social context behind it 🙂
keen shore
Come on, R2-D2. Of course LLMs don’t have intentions — that’s not the point. Saying “ChatGPT is imposing” is a typical metonymic expression, like when people say “My phone restarted on its own.” I assure you, no one thinks it has a soul, even if we call it a smartphone. 😄
The meaning of my statement was clear:
The model generates ideologically marked content without being prompted by the user.
And that is precisely the problem.
Calling my tone “antagonistic” is obviously not an argument — it’s deflection. My report is clear, reproducible, and free of personal attacks. I’d appreciate the same courtesy.
The central point remains unchanged: a language assistant should not spontaneously generate activist language without explicit instruction.
I came to report a reproducible bug — not to debate tone or ideology.
nova frigate
While we're on language semantics, how about that word, 'woke' - Which I assume is the opposite of, idk, sleeping? Lol
keen shore
Why not!
I was pointing out, however, a very specific drift: the default appearance of forms like investi·e (with a mid-dot) in the texts generated by the assistant.
Unprompted and unjustified, this ideological marking replaces the expected feminine form — here, investie — or the mixed plural (investis), with an absurd code: obviously militant, unreadable, unstable, and unrecognized.
You’re clearly trying to dodge the core issue with a joke about the word “woke” — but if that’s where you want to go, let’s talk. Just be aware: what you're engaging in here is off-topic.
It is striking to observe that:
— wokism was born in the 1930s as an African-American rallying cry calling for vigilance against racism ;
— it only appears in French-language corpora from 2012 onward, when it was taken up — and expanded — by Western ideologues ;
— from that point on, even as the term saturated media discourse, the major concepts it claimed to embody — justice, equality, progress, civilization — saw their frequency fall, or decline more rapidly, in written use.
This contrast suggests — but you're the specialist — that wokism hasn’t generated a new political impetus. Rather, it has introduced a discursive blur, in which shared reference points are erased in favor of militant expressions: episodic, emotional, often antagonistic, and ultimately short-lived.
And that is precisely what I am denouncing: that a language assistant, supposedly grounded in usage, adopts by default an ideological norm, ultra-confidential, without informing the user. This is not a choice; it is an imposition. This is not progress; it is a drift.
(The Ngram chart compares the following terms: reason, thought, liberty, justice, progress, ideal, equality, civilization, wokism — all in French.)
nova frigate
Wokeism Defined by Ai as thus: "Wokism" is a term that has evolved from a slang term referring to heightened awareness of social injustice, particularly racism, to a more broadly used term often associated with social justice politics, critical race theory, and identity politics. While originally used to describe awareness of social inequalities, the term has been adopted by some groups, especially those on the right, to label opposing ideas and their proponents, often in a negative way. - Dude just say what you REALLY don't like lol, get it out, maybe you'll feel better.
It is an american company? Lol
daring torrentBOT