Being accused of using writing software

Don't be too hasty. Maybe O.P. hasn't seen any of your suggestions yet. Either way, you went out of your way to help, and the rest is on them.

On topic, we often debate that leaving grammar and stylistic errors helps? AI supposedly doesn't make them, and it shouldn't mix tenses the way the O.P. did? Or does the algorithm only take into account each sentence individually, so the mixing of tenses doesn't come into play?
I asked a friend:

1️⃣ AI, Grammar, and Stylistic Errors
  • AI models (including me) are designed to avoid grammar and stylistic errors unless asked to mimic a particular voice.
  • Humans often debate whether leaving “natural” errors makes writing more authentic. In practice, consistent grammar usually improves clarity, especially for factual or analytical writing.

2️⃣ Mixing Tenses: Should It Happen?
  • AI is not supposed to mix tenses arbitrarily the way human writers sometimes do.
  • When AI does mix tenses unintentionally, it is typically a generation inconsistency that ideally should be edited out.
  • That said, deliberate tense shifts (e.g., moving from past to present for immediacy) are acceptable if done intentionally.

3️⃣ Does the Algorithm Consider Only Each Sentence Individually?
  • No, large language models (LLMs) like GPT do not process each sentence in isolation; they consider context windows (thousands of tokens) to maintain coherence, tense consistency, and logical flow.
  • However, under certain conditions (prompt ambiguity, cut-off windows, or fast generation), models may lose track of tense consistency across sentences, leading to the kind of mixing you noticed.

In summary:
  • Leaving errors for “authenticity” may be valuable in creative or conversational writing but typically degrades clarity in analytical work.
  • AI should maintain tense consistency unless intentionally shifting.
  • The algorithm takes context into account, but imperfect tracking can still cause tense mixing, which should be corrected during review.
 
I've edited a few AI-generated translations that were inconsistent in their use of tense and contained typos and grammatical mistakes. Of course translations also draw on translation memory databases, with all their imperfections, so that adds to the muddle.
 
I asked a friend:

1️⃣ AI, Grammar, and Stylistic Errors
  • AI models (including me) are designed to avoid grammar and stylistic errors unless asked to mimic a particular voice.
  • Humans often debate whether leaving “natural” errors makes writing more authentic. In practice, consistent grammar usually improves clarity, especially for factual or analytical writing.

2️⃣ Mixing Tenses: Should It Happen?
  • AI is not supposed to mix tenses arbitrarily the way human writers sometimes do.
  • When AI does mix tenses unintentionally, it is typically a generation inconsistency that ideally should be edited out.
  • That said, deliberate tense shifts (e.g., moving from past to present for immediacy) are acceptable if done intentionally.

3️⃣ Does the Algorithm Consider Only Each Sentence Individually?
  • No, large language models (LLMs) like GPT do not process each sentence in isolation; they consider context windows (thousands of tokens) to maintain coherence, tense consistency, and logical flow.
  • However, under certain conditions (prompt ambiguity, cut-off windows, or fast generation), models may lose track of tense consistency across sentences, leading to the kind of mixing you noticed.

In summary:
  • Leaving errors for “authenticity” may be valuable in creative or conversational writing but typically degrades clarity in analytical work.
  • AI should maintain tense consistency unless intentionally shifting.
  • The algorithm takes context into account, but imperfect tracking can still cause tense mixing, which should be corrected during review.
Not so sure about #3 though.

The way many of these AI detectors seem to work is that they always underline isolated sentences within the text as suspicious. It's probably the algorithm working as intended, as I guess many people try to hide their usage of AI by mixing AI-generated sentences with their own.
 
... and breathe.

Pros:

Passable proto-rant.

Cons:

Requires more expletives and imagination.
Overuse of capitalization hinders score.
Performed in incorrect location before jaded audience.
Clearly emotional rather than heartfelt.

Final score: 3/10

Advice for ranter: Think about the effect you wish your rant to have, and try to channel that into your next rant.

Add our livers to the con list

Drink, everyone.
 
I've edited a few AI-generated translations that were inconsistent in their use of tense and contained typos and grammatical mistakes. Of course translations also draw on translation memory databases, with all their imperfections, so that adds to the muddle.
It could be that ChatGPT is sometimes wrong, of course 😱
 
I'm beginning to understand why the Underground Lesbian Army's assault on society has stalled. All you lot do is sit around and smirk at double entendres.

"Hetty, would you read the minutes of the last meeting please?"
"I can't remember the last time I had a meating. Nobody's meated me for years."
"Hihihi. Nice one. We still have to plan our uprising though."
"What are we, men? Nothing needs to rise up."
Etc.
 
I'm beginning to understand why the Underground Lesbian Army's assault on society has stalled. All you lot do is sit around and smirk at double entendres.
We have a secret weapon, though - we can multitask. Just because we're planning to overthrow the world with sensible leggings and comfortable shoes doesn't mean we can't have fun while doing it. Just because you're not seeing the results doesn't mean they aren't there. We're like erosion. Nothing to see on a day to day basis but come back in a year and your wives and daughters are stylish, erudite, and linguistically adroit... oh, and your wife's new bestie is Emma from the ULA's outreach program.

Don't blink. We know where you sleep. muahahahahahaha
 
We have a secret weapon, though - we can multitask. Just because we're planning to overthrow the world with sensible leggings and comfortable shoes doesn't mean we can't have fun while doing it. Just because you're not seeing the results doesn't mean they aren't there. We're like erosion. Nothing to see on a day to day basis but come back in a year and your wives and daughters are stylish, erudite, and linguistically adroit... oh, and your wife's new bestie is Emma from the ULA's outreach program.
My wife has always been stylish, erudite and linguistically adroit. The sensible leggings and comfortable shoes are new, though.
Don't blink. We know where you sleep. muahahahahahaha
Help, help, there are lesbians watching me sleep!
 
we're planning to overthrow the world with sensible leggings and comfortable shoes doesn't mean we can't have fun while doing it.
If we could get pants that have actual, real functioning pockets like guys get, I would be fine with giving supreme military power to the ULA.
 
If we could get pants that have actual, real functioning pockets like guys get, I would be fine with giving supreme military power to the ULA.
Sometimes I'm very lucky and find hiking trousers at Mountain Warehouse that have multiple pockets. Multiple!
 
We have a secret weapon, though - we can multitask. Just because we're planning to overthrow the world with sensible leggings and comfortable shoes doesn't mean we can't have fun while doing it. Just because you're not seeing the results doesn't mean they aren't there. We're like erosion. Nothing to see on a day to day basis but come back in a year and your wives and daughters are stylish, erudite, and linguistically adroit... oh, and your wife's new bestie is Emma from the ULA's outreach program.

Don't blink. We know where you sleep. muahahahahahaha
Do you have to be a spelunker to join the ULA, or just to enjoy going down… underground I mean, of course?
 
Back
Top