**Algorithmic Law - Project/ You said: I want to focus on the first part of the site.Signs of AI Writing (standalone src).htmlhtml** I want to focus on the first part of the site. Signs of AI writings. It A SOudns like AI itself and B doesnt cover the signs of AI writing that wikipedia talks about and really boils it down too much https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing#Promotional_and_advertisement-like_language Let's have a section called Undue Emphasis on Significance and NotabilityApr 21 **Claude responded: Let me look at the current HTML file to see exactly how the signs are structured, so I can add the new section to match the existing pattern.**Mapping article structure with sign componentsMapping article structure with sign components Let me look at the current HTML file to see exactly how the signs are structured, so I can add the new section to match the existing pattern.Architected new content section matching existing site structureArchitected new content section matching existing site structure You're right on both counts. The Wikipedia page covers significance-stuffing and notability-laundering as distinct phenomena from vague puffery, and they're arguably the single most AI-tell on modern outputs — the others can be prompted away, but this one is baked into the register the model has learned for "important-sounding prose." Here's a draft of sign 07 that tries to do two things at once: faithfully split the Wikipedia distinction into its two flavors, and stay out of the tics the site itself catalogs. Drop it in after sign 06 (Uniform Rhythm), just before the "A caveat" section-head.html
`<article class="sign"> <div class="sign-number">07</div> <h2 class="sign-title">Undue Emphasis on Significance and Notability</h2> <div class="sign-body"> <p>Every subject, no matter how minor, gets described as pivotal, enduring, a watershed. A 1989 regional statistics office becomes "a pivotal moment in the evolution of regional statistics." A local restaurant's passing mention in the news becomes coverage "underscoring its role as a well-known late-night venue." Someone interviewed on three podcasts has been "profiled in multiple high-quality, independent, and widely-read outlets." The model has learned the register of textbook-history and encyclopedia-lede writing, and now applies it to a corner grocery.</p>
<p>This splits into two related tells, worth keeping apart because they deploy differently.</p>
<p><strong>Significance-stuffing.</strong> Sweeping claims about legacy, impact, or broader trends get attached to specific local facts, usually through a present-participle phrase tacked onto the end of the sentence: <em>…marking a pivotal moment… …reflecting broader shifts… …contributing to the enduring legacy of…</em> The grammar does the work of suggesting the content is profound. The content isn't.</p>
<div class="example machine">
<span class="example-label">Machine</span>
"The establishment of the town's post office in 1892 marked a pivotal moment, reflecting broader trends in rural American infrastructure development and contributing to the enduring legacy of frontier connectivity."
</div>
<p><strong>Notability-laundering.</strong> Lists of outlets, awards, and credentials get piled up as if their accumulation were substance. A sentence names four magazines, says nothing about what any of them actually wrote, and ends with a flourish about the subject being a "leading voice." The wording often echoes the exact language of the venue the model thinks it's writing for — Wikipedia drafts get "independent coverage" and "significant, substantial, secondary coverage"; corporate bios get "widely recognized industry leader"; press kits get "featured in."</p>
<div class="example machine">
<span class="example-label">Machine</span>
"Her insights have been featured in <em>Wired</em>, <em>Refinery29</em>, and other prominent media outlets, underscoring her role as a leading voice in the emerging conversation around digital privacy."
</div>
<p><strong>Why it happens.</strong> Training data is full of this register used <em>correctly</em> — textbook summaries of real historical milestones, obituaries of genuinely consequential people, Wikipedia ledes about actual turning points. The model has learned the register without learning when it's earned. Meanwhile, regression to the mean does its work in the other direction: a specific, unusual fact ("inventor of the first train-coupling device") gets smoothed upward into the generic and flattering ("a revolutionary titan of industry"). The portrait blurs while the caption grows louder.</p>
<p><strong>How to spot it.</strong> Apply the replaceable-subject test: does the sentence still make sense if you swap the noun? "X marked a pivotal moment, reflecting broader shifts in the evolving landscape of Y" works for almost any X and Y. And watch for attribution pile-ups — four outlets named in a row without a word about what any of them said is performing notability, not reporting it.</p>
<p><strong>How to fix it.</strong> Don't assert significance; demonstrate it. If the thing is important, the specific detail that makes it important is already interesting. "Opened in 1892" is a fact. "A pivotal moment reflecting broader trends" is a press release. One of those tells the reader something.</p>
</div>
</article> One small companion edit: the intro on line 45 says "Six tells." Change to "Seven tells." A couple of notes on the broader concerns you raised: On the site sounding AI-like — the thing I'd actually point at is the rigid **Why it happens / How to spot it / How to fix it** scaffold repeated identically across every sign. That's the exact "rigid outline structure" Wikipedia calls out. Some signs could break the pattern — sign 06 (Uniform Rhythm) is already close to doing this. I kept the scaffold in sign 07 for consistency, but if you want to loosen the format site-wide, that would go furthest toward making the page sound less like what it's describing. On the "boils it down too much" concern — the Wikipedia page has at least four other categories the site doesn't touch: **Superficial analyses** (the vague-attribution "Critics have praised…" move), **Knowledge-cutoff disclaimers**, **Outline-like "Despite these challenges…" conclusions**, and the whole **Markdown-everywhere** problem. If you want, I can draft any of those in the same style. The outline-like-conclusions one in particular pairs nicely with sign 05 (Compulsive Wrap-Up) — it's the structural version of what sign 05 describes at the sentence level. **You said: Dont do tells just says signs, can you show me in ehre** Dont do tells just says signs, can you show me in ehreApr 21 **Claude responded: Found six spots.**Identified multiple instances and distinguished verb from noun usageIdentified multiple instances and distinguished verb from noun usage Found six spots. Five are the noun "tell/tells" (the AI-writing concept) and should change. One is the verb "to tell" (as in "perceive"), which is a different word — I'd leave it alone, but flagging it so you can decide. **Change these five:** • **Line 45** (the lede under "The signs"): Six tells. → Seven signs. *(already updating the count for sign 07)* • **Line 62** (example label in sign 01): A bouquet of tells → A bouquet of signs • **Line 68** (in sign 01, "How to spot it"): is almost always a tell → is almost always a sign • **Line 182** (the coda at the end of the Signs section): The tells shift. → The signs shift. now the tell is something subtler → now the sign is something subtler • **Line 307** (in the "Will we all sound like this?" tab): The tells also generate an immune response → The signs also generate an immune response **Leave this one:** • **Line 199** (opener of the "Why" tab): You can tell.` — Here "tell" is a verb meaning perceive / figure out, not the noun. Reads naturally; rewording would flatten it. Say the word, though, if you want it gone anyway and I'll find a replacement.
Want me to also go through the new sign 07 draft I gave you and strip any "tell" language there? I think I used the word "tells" once in the transition sentence ("This splits into two related tells…") — that would become "…two related signs."
You said: no i meant can you create the specifc signs of AI into an artifcat i can see here
no i meant can you create the specifc signs of AI into an artifcat i can see hereApr 21
**Claude responded: Got it — I'll build a standalone artifact that renders sign 07 exactly as it would look on the page.**Deliberated creating HTML artifact matching existing sign aestheticDeliberated creating HTML artifact matching existing sign aesthetic
Got it — I'll build a standalone artifact that renders sign 07 exactly as it would look on the page. Let me put it together.Viewed a file, created a file, read a fileViewed a file, created a file, read a file
Here's sign 07 rendered in the site's own aesthetic — same fonts, same cream palette, same cold-steel "Machine" example rule, same sign number in italic Fraunces. What you're seeing is exactly what it'll look like when you paste the HTML into place after sign 06.
A couple of things I was deliberate about:
The bold inline labels (Significance-stuffing, Notability-laundering) split the sign into its two Wikipedia-faithful sub-types without needing sub-headings that would break the visual rhythm of the other signs. Each sub-type gets its own machine example directly underneath, so you can see the pattern.
I left off a sign-tag (the colored dot + category label at the top) because this one isn't used in the quiz, matching signs 05 and 06. If you do want to add a tag for category grouping — something like "Puffery · structural" to relate it to sign 03 — say the word and I'll add it.
If you want, I can also produce the other missing Wikipedia categories as additional preview artifacts: Superficial analyses, Outline-like "Despite these challenges" conclusions, or Knowledge-cutoff disclaimers are the three that would most meaningfully round out the taxonomy.Sign 07 previewCode · HTML
Actually, I was talking about putting undue emphasis on this slide, the signs. Let's talk, what do you think is better, putting all the signs on one page or on separate areas? Lets say the signs of AI writing has a quick blurb then you can click on it to go to deeper analysis and example and suchClaude is AI and can make mistakes. Please double-check responses.
Six tells. → Seven signs. (already updating the count for sign 07)A bouquet of tells → A bouquet of signsis almost always a tell → is almost always a signThe tells shift. → The signs shift. now the tell is something subtler → now the sign is something subtlerThe tells also generate an immune response → The signs also generate an immune responseYou can tell. — Here "tell" is a verb meaning perceive / figure out, not the noun. Reads naturally; rewording would flatten it. Say the word, though, if you want it gone anyway and I'll find a replacement.