Various LLM Smells
by speckx on 5/28/2026, 7:02:18 PM
https://shvbsle.in/various-llm-smells/
Comments
by: Planktonne
> The LLM generated writing obviously felt significantly better than my own writing.<p>A general pattern for LLMs is that they look really good at things you are bad at. What that means is that if you find yourself thinking of its output as <i>significantly</i> better than yours in a particular domain, there's a high chance that you are not equipped to judge that quality effectively.
5/28/2026, 8:53:54 PM
by: tptacek
The LLM writing sameness is bad. Use LLMs to help your writing! But don't include a word they generate, even just a vocabulary adjustment, in your own output. Have them critique structure and flow, spot overused words and passive constructions and dumb picks for topic sentences. It's great for that, and those are all objective improvements in your writing that won't mess up your style.<p>The LLM sameness in web design is good. Most sites shouldn't try to be idiosyncratic. The best design for a site with real utility is <i>legibility</i>, and LLMs are better at that than the median developer. Always laying out the same buttons? Always using the same type scales? Good! If it looks good to you, you weren't going to do better on your own, and you were very likely to do worse.
5/28/2026, 9:54:57 PM
by: spdustin
- “(The) honest caveat:” (or “genuine caveat:”, both with the colon)<p>- “(The) honest answer:” (again, with colon)<p>- “The thing to internalize:”<p>- “The smoking gun:”<p>(really, sentences that start with “The <tag suggesting the next clause is the key point>:” are a strong tell, but those four are the most prolific)<p>- “load bearing” (when not talking about architecture)<p>- “blast radius” (when not talking about actual explosives, but rather the effect of an event/action)<p>- “smoke test” (esp. when “sanity check” is more apropos)<p>- Lists of three clauses/adjectives where the third is really just a combination of the first two<p>- Referring to the “shape” of things figuratively<p>- Social media posts that end with “Curious if anyone…”<p>- Stories or anecdotes using. “Oh. Oh.” (where the second “oh” is italicized)<p>Edit: Yes, some of those last ones are terms that we often use as devs...but I would argue about the actual frequency of their use. Plus, these tells live on in prose generated by the latest models.
5/28/2026, 9:13:35 PM
by: rimeice
Scrolling down a LinkedIn feed is hilarious at the moment.<p>My favourite one today from today:<p>“The tax isn't the problem. The mindset is.”
5/28/2026, 9:15:22 PM
by: Hfuffzehn
The interesting thing for me is that I do not feel like the writing of LLMs has improved very much lately stylistically. They have reached a "good" level some time ago but the newer models havn't brought such improvements that you would prefer them to an expert human writer.<p>Will be interesting if that holds in other areas when chasing super intelligence.
5/28/2026, 10:02:58 PM
by: n42
<p><pre><code> No ___, no ____. Just _____ </code></pre> or using "honest" to describe an approach.
5/28/2026, 8:40:59 PM
by: kylemaxwell
At this point, I want somebody's raw(ish) writing, with spelling errors and grammar mistakes and whatever, at least when it comes to most writing: blog posts, Slack messages, etc. LLMs are great for helping generate ideas, writing code, and maybe even cleaning up some writing, but doing the writing overall? Please don't. I want to hear what you have to say, not what the AI says, if it's something along those lines.
5/28/2026, 10:01:12 PM
by: KronisLV
> The "JetBrains Mono" font<p>Thought for sure we'd get a critique of Inter overuse. JetBrains Mono is a lovely font, though.
5/28/2026, 8:51:21 PM
by: 1970-01-01
The LLM doesn't smell like authentic writing but it does a great job for fast and cheap words. We've gained something similar to fast food. Words made very cheap, very fast, easily digestible, but they have no emotion. In short stints it does have a place in the world.
5/28/2026, 9:19:52 PM
by: newer_vienna
I don't think I've met anyone who uses the word "genuinely" as much as Claude does.
5/28/2026, 10:01:21 PM
by: nijave
> :black_circle_for_record: Smoking gun.<p>> "belt and suspenders"
5/28/2026, 9:46:09 PM
by: dvt
It's kind of interesting how genuinely <i>hard</i> it is to get models to deviate from basically all of these tropes. You can straight up tell it "I hate that card design, do something different, get creative!" and it'll do something either (a) ugly as sin (clearly just essentially a random walk through parameters) or (b) some same-y derivation of that card.<p>In coding, I've noticed a few tropes as well: everything is a "contract" or an "artifact" (clearly trained on like three decades of Java lol), everything is constantly "backwards-compatible" or "versioned" (even if working on a brand new greenfield project), and a few others.
5/28/2026, 8:51:22 PM
by: docheinestages
You are right to push back.
5/28/2026, 9:06:20 PM
by: newer_vienna
Thank you, these are all things I've noticed too.
5/28/2026, 9:59:14 PM
by: danielodievich
All of those are included in the bulk of the documents passing my work input these days. It is infuriating. Out of principle I maintain 100% me in all my writing but I don't know if it matters. Well maybe it does... an interviewee recently complimented me on the "nicest and most human resume" they saw recently. That felt good
5/28/2026, 8:48:07 PM
by: viccis
One Python one I hate is that it adds crazy amounts of newlines for no real readability gains.<p>Instead of this:<p><pre><code> def add_three_ints(x: int, y: int, z: int) -> int: return x + y + z </code></pre> it will write:<p><pre><code> def add_three_ints( x: int, y: int, x: int, ): return x + y + z </code></pre> While it's always preferable to do this when you get either long or complex function signatures, Opus 4.7 and GPT 5.5 do this everywhere. When you combine it with their penchant for writing helper functions for everything, you get a ton of vertical padding that messes up the readability imo because Python really relies on your eye seeing indents for scope.
5/28/2026, 10:17:13 PM
by: antoineMoPa
Abusing the words "canonical" and "normalized".
5/28/2026, 9:44:23 PM
by: speak_plainly
I came here for the performative anti-AI intellectualism and was not disappointed.
5/28/2026, 10:00:51 PM
by: TacticalCoder
So the year is 2026 and we cannot point a LLM at, say, this HN thread, and give it the instructions: <i>"I don't want to look like a dumbass, so don't make these obvious mistakes / don't use these obvious tells"</i>!?
5/28/2026, 10:01:34 PM
by: mil22
Those cards, so familiar! Exactly what Opus produced for me.<p>Did Anthropic and/or OpenAI deliberately train their models to produce websites with a specific design language, or did these stylistic preferences emerge naturally as some kind of LLM-selected optimum?
5/28/2026, 8:52:47 PM
by: barrkel
Quietly. Clean. Honest. Sharp take.
5/28/2026, 10:07:18 PM
by: manoDev
Welcome to the future of fast-food software. Taste of deep frying and preservatives.
5/28/2026, 9:06:36 PM
by: dionian
KPI cards, purple gradients
5/28/2026, 8:49:11 PM
by: poszlem
What I find amazing is how HARD it is to make the LLM produce a piece of text that does not sound like slop. I have had dozens of sessions where I tried to make it write like a human would, and yet it still uses those tired writing phrases. I don't understand why neither openai, nor anthropic are able to do anything to make it better, and in some cases it feels like we are actually going backwards.
5/28/2026, 9:12:41 PM
by: stanleydupreez
[flagged]
5/28/2026, 10:09:06 PM
by: willyv3
[flagged]
5/28/2026, 9:48:44 PM
by: nikhilpareek13
[dead]
5/28/2026, 9:29:37 PM