Humanize AI Text: Bikin Tulisan Lolos Detektor dengan Skor 0% AI

Wikipedia punya halaman namanya "Signs of AI writing" yang nyatet tiap pola yang dicari AI detector. Kita bakal kasih seluruh list itu ke Claude, suruh dia nggak pernah ngelakuin hal-hal itu, dan simpen jadi skill reusable.

Skill itu set instruksi tersimpan yang Claude ikutin tiap kali lu aktifin. Pas lu ketik /human, Claude masuk ke mode humanizer dan nerapin semua aturannya otomatis. Lu bangun sekali, install, dan dia jalan selamanya.

Buka Claude dan mulai chat baru. Paste prompt ini:

I want you to humanize all my text moving forward so it doesn't sound like AI.
Go to this Wikipedia page:

en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing

Read through every single pattern listed on that page. From now on, never use any of those patterns in your writing. List out every pattern you found so I can confirm you understand them all.

Claude bakal baca halaman Wikipedia-nya, narik semua pola, dan ngelistnya balik ke lu. Cek listnya buat mastiin nggak ada yang kelewat.

Tip: Lu nggak perlu copy-paste halaman Wikipedia-nya sendiri. Claude baca langsung pas lu kasih URL-nya.

Claude harusnya balik dengan sekitar 24 pola di empat kategori:

• Content patterns: inflated significance, promotional language, vague attributions
• Language patterns: AI vocabulary, copula avoidance, synonym cycling
• Style patterns: em dash overuse, excessive bold, inline header lists
• Communication patterns: chatbot pleasantries, sycophantic tone, filler phrases

Kalau ada yang kelewat, suruh Claude balik dan lihat lagi.

Tetap di chat yang sama, ketik:

/skill-creator

Build me a personal humanizer skill based off everything we just found so I never have to explain any of this ever again. Name it 'human'. Make it activate when I type /human.

Claude bakal bikin file skill lengkap dengan semua aturannya udah ke-bake di dalam.

Setelah Claude generate skill-nya, klik install. Skill-nya ke-tambah ke setup Claude lu secara permanen.

Buat makainya, tinggal ketik /human di awal chat mana pun. Terus nulis kayak biasa. Claude otomatis nerapin semua aturan humanizer.

Penting: Tetap di chat yang sama buat keempat step. Kalau lu mulai chat baru di tengah-tengah, Claude kehilangan konteks pola Wikipedia-nya.

• Inflated significance: "Serves as a testament to..." "Pivotal moment in the evolution of..." Fix: langsung bilang apa itu, skip puffing pentingnya.
• Undue notability claims: "Featured in The New York Times, BBC, and Wired..." Fix: kutip satu sumber dengan detail spesifik.
• -ing phrase padding: "Highlighting... Underscoring... Emphasizing..." Fix: akhirin kalimatnya, mulai yang baru.
• Promotional language: "Vibrant" "Nestled" "Breathtaking" "Groundbreaking" Fix: pakai deskripsi plain dan spesifik.
• Vague attributions: "Experts say..." "Industry observers note..." Fix: sebut sumber aslinya atau buang klaimnya.
• Challenges and prospects: "Despite challenges... the future looks bright" Fix: sebut fakta spesifik, skip optimismenya.

• AI vocabulary words: "Delve" "Crucial" "Landscape" "Tapestry" "Foster" "Leverage" Fix: pakai kata normal. "Look into" bukan "delve".
• Copula avoidance: "Serves as" "Stands as" "Boasts" ganti "is" atau "has" Fix: langsung bilang "is" atau "has".
• Negative parallelisms: "It's not just X, it's Y" "Not only... but also..." Fix: bilang poinnya langsung.
• Rule of three abuse: "Innovation, inspiration, and industry insights" Fix: jangan maksa tiga item, dua atau empat juga fine.
• Synonym cycling: "The protagonist... the main character... the central figure..." Fix: pilih satu nama, stick ke situ.
• False ranges: "From X to Y" di mana X dan Y nggak di skala nyata Fix: list topiknya, skip framing range-nya.

• Em dash overuse: pakai em dash di mana-mana buat efek dramatis Fix: pakai koma, titik, atau kurung.
• Excessive bold: tiap frase di-bold buat penekanan Fix: bold seperlunya atau nggak sama sekali.
• Inline header lists: "Speed: The speed has improved..." Fix: nulis dalam prosa, skip bold label-nya.
• Title case headings: "Strategic Negotiations And Global Partnerships" Fix: sentence case: "Strategic negotiations".
• Emoji decoration: bullet point mulai dengan emoji Fix: buang emoji kecuali audiens-nya emang ngarep.
• Curly quotes: pakai curly quotes ganti straight quotes Fix: pakai straight quotes di plain text.

• Chat artifacts: "I hope this helps!" "Let me know if..." "Certainly!" Fix: buang semua basa-basi chatbot.
• Cutoff disclaimers: "As of my last update..." "Based on available info..." Fix: sebut fakta langsung tanpa disclaimer.
• Sycophantic tone: "Great question!" "You're absolutely right!" Fix: skip validasinya, jawab pertanyaannya.
• Filler phrases: "In order to" "Due to the fact that" "At this point in time" Fix: "To" "Because" "Now".
• Excessive hedging: "Could potentially possibly be argued that it might..." Fix: pilih posisi. "The policy may affect outcomes."
• Generic conclusions: "The future looks bright" "Exciting times ahead" Fix: akhirin dengan fakta spesifik atau next step.

Contoh 1: Blog Intro

Before (kedengeran AI):

In today's rapidly evolving technological landscape, AI-assisted coding serves as an enduring testament to the transformative potential of large language models, marking a pivotal moment in the evolution of software development.

After (kedengeran manusia):

AI coding assistants speed up some tasks. In a 2024 Google study, developers using Codex finished simple functions 55% faster but showed no improvement on debugging or architectural decisions.

Teks before nampol lima AI tell dalam satu kalimat. Teks after nyebut fakta spesifik dengan sumber. Itu bedanya.

Contoh 2: Product Description

Before (kedengeran AI):

This groundbreaking tool boasts a vibrant interface, offering a seamless experience that fosters collaboration and enhances productivity. It's not just a tool -- it's a catalyst for innovation.

After (kedengeran manusia):

The app has a shared workspace where your team can edit the same document at the same time. It saves about 2 hours per week compared to emailing files back and forth.

Teks before pakai kata promosi dan negative parallelism. Teks after ngedeskripsiin apa yang beneran dia lakuin dengan angka penghematan waktu yang spesifik.

Begitu humanizer skill lu ke-install, pasangin sama prompt-prompt ini.

Match a Specific Voice

/human

Here's an example of my writing style: [paste 2-3 paragraphs of your own writing here]. Match this voice, sentence rhythm, and vocabulary level for everything you write in this chat.

Rewrite Existing AI Text

/human

Rewrite the following text so it reads as 100% human-written. Remove every AI pattern. Keep the same information but change the sentence structures, word choices, and rhythm.

[paste your AI text here]

Academic Humanizer

/human

Write this in an academic tone but still avoid all AI patterns. Use formal language without the filler. Cite sources properly. No hedging, no sycophancy, no promotional language. Topic: [your topic]

Read the text below and identify every sign of AI writing. List each pattern you find with the specific phrase that triggered it. Then rewrite the text with all patterns removed.

[paste text to audit]

Jalanin ini di tulisan lu sendiri sebelum publish. Bahkan dengan humanizer skill aktif, kadang ada pola yang lolos. Audit cepat nangkep apa pun yang skill-nya lewatin.

• Perplexity: seberapa ketebak kata berikutnya? AI cenderung milih kata paling mungkin. Manusia lebih nggak ketebak. Skill lu handle ini dengan variasiin struktur kalimat dan ngehindarin pilihan kata AI paling umum.
• Burstiness: manusia nulis dengan ritme nggak rata. Ada kalimat pendek. Ada yang panjang berklausa banyak. AI nulis dengan panjang kalimat konsisten. Skill lu benerin ini dengan ngaduk ritme secara sengaja.
• Vocabulary distribution: AI overuse kata tertentu. "Delve," "crucial," "landscape," "tapestry," dan "foster" muncul jauh lebih sering di teks AI. Skill lu ganti ini dengan alternatif normal.

Heads up: Nggak ada metode yang 100% reliable buat ngelewatin AI detector di semua situasi. Detektor terus update. Tujuannya nulis teks yang beneran kedengeran manusia, bukan ngakalin satu tool spesifik. Kalau tulisan lu kebaca natural sama manusia, dia bakal lolos kebanyakan detektor kebanyakan waktu.