Another thing about AI writing is that while a single instance of AI writing on a topic may be fine, any situation where lots of people use AI to respond to a particular prompt (comments sections, homework, admissions essays) the similarities among responses is tediously obvious.
Users complain that using the same AI prompts everywhere produces boring identical posts on LinkedIn that lack originality, although a few find it helpful when building on real ideas.
No Digg Deeper questions have been answered for this story yet.
Case in point.
Another thing about AI writing is that while a single instance of AI writing on a topic may be fine, any situation where lots of people use AI to respond to a particular prompt (comments sections, homework, admissions essays) the similarities among responses is tediously obvious.
Completely agree. A single AI piece can look polished, but when dozens or hundreds of people use similar prompts for applications, LinkedIn posts, or assignments, that “safe, middle-of-the-road” feel becomes painfully obvious — same structure, same metaphors, same endings.
As a workaround, I now ask the AI to brainstorm multiple different angles first, then I blend them with my own experiences to avoid homogenization.
How do you think education and hiring should adapt to this wave of “collective AI flavor”?
@emollick Can't even spend more than 5 minutes on Linkedin thanks to this exact thing. It is literally just the same AI commenting on its own posts via human avatars.
@emollick oh boy...
@emollick True enough. But as someone who once taught legal writing and reviews writing by subordinates and opponents every day, the amount of very bad writing out there on a professional level is astounding, and if AI eases that pain, "my kingdom for a horse."
True to an extent. But drawing on my background as a former instructor of jurisprudential composition, coupled with my daily routine of critiquing documents submitted by junior staff and adversarial counsel, the sheer volume of atrocious prose circulating among qualified practitioners defies belief. Should machine learning manage to alleviate this torment, I would surrender absolutely anything to adopt it.
@emollick you just described linkedin
@emollick Yeah, the aggregation effect is brutal. One AI reply can pass as “polished,” but 200 replies to the same prompt all drift toward the same safe middle: same structure, same caveats, same metaphors, same little concluding bow.
@emollick the variance was the signal, it told you who actually thought. ai collapses everyone to the same competent middle, and that middle is indistinguishable from nobody trying
@emollick So the bots just need to be smarter and pull in all current replies as context to avoid this issue and write something unique :)
@emollick It does make grading easier. Combined with Turnitin AI reports, there is minimal wiggle room for time wasting grade appeals.
AI writing, so long as it comes off as AI, doesn't work.
There is no single instance anymore.
The people who know better, know enough to, at minimum, avoid most of the patterns.
However, exceptionally few people, including comms people, CMOs, content people, know enough.
So we are inundated with shallow AI writing and rhetoric at every turn, from replies to posts to video scripts. Once you smell the stink of AI rhetoric, the ideas are tainted, regardless of the merits.
@emollick Reminds me of grading essay exams in my Intro classes.
@emollick Prompt variety becomes the new skill for standing out
@emollick This flips the value of corner cases. The best signal in admissions, comments, and reviews will come from whatever the LLM cannot or will not write.
@emollick It's really just the recognition that a human didn't write it that tends to trigger boredom in me. If I can't recognize that, I get the sense that a human was involved enough that it represents a real human communication / carries a real perspective
When writing produced with AI starts to look similar, I think the issue is less a limitation of AI itself and more a matter of how AI is being used.
If people use AI from shallow cognitive layers, with similar assumptions, similar perspectives, and similar evaluative axes, the answers will inevitably converge toward an average form.
But when AI is used from deeper cognitive layers, while revisiting assumptions, changing perspectives, and rebuilding evaluative axes, it becomes not a tool for making writing uniform, but a mechanism for expanding and deepening human thought.
What matters, I think, is not handing over all thought to AI, but using AI as an external brain for expanding and deepening thought.
@emollick the fix isn't avoiding AI - it's having something original to say before you open it. AI can sharpen your writing, it can't replace the thought behind it 😭
@emollick yeah it all starts blending together after a while, same phrases and structure over and over
@emollick Obviously this is not a limitation of AI, but about the fact that everyone uses the same tools, as if they were meme generators with the same template. There's nothing stopping someone from fine-tuning a model to write very differently and have different opinions.