跳到主要內容

Humanizer

人性化文本:去除 AI 腔調並增添真實語氣。

技能元數據

來源捆綁(默認安裝)
路徑skills/creative/humanizer
版本2.5.1
作者Siqi Chen (@blader, https://github.com/blader/humanizer),由 Hermes Agent 移植
許可證MIT
平臺linux, macos, windows
標籤writing, editing, humanize, anti-ai-slop, voice, prose, text
相關技能songwriting-and-ai-music

參考:完整 SKILL.md

信息

以下是 Hermes 在觸發此技能時加載的完整技能定義。這是技能激活時代理看到的指令。

Humanizer:去除 AI 寫作模式

識別並去除 AI 生成文本的痕跡,使寫作聽起來自然且像出自人類之手。基於維基百科的“AI 寫作跡象”指南(由 WikiProject AI Cleanup 維護),該指南衍生自對數千個 AI 生成文本實例的觀察。

核心洞察: 大語言模型(LLM)使用統計算法來猜測接下來應該出現什麼。結果往往傾向於統計上最可能的補全內容,這正是以下典型模式被固化的原因。

何時使用此技能

當用戶要求執行以下操作時,加載此技能:

  • “人性化”、“去 AI 化”、“去水詞”或“去 ChatGPT 味”一段文本
  • 重寫某些內容,使其聽起來不像由 LLM 編寫
  • 編輯草稿(博客文章、論文、PR 描述、文檔、備忘錄、電子郵件、推文、簡歷要點),使其聽起來更自然
  • 在其生成的寫作中匹配他們的語氣
  • 在發佈前審查文本中的 AI 痕跡

此外,在編寫面向用戶的散文(如發佈說明、PR 描述、文檔、長篇解釋、摘要)時,也將此技能應用於你自己的輸出。Hermes 的基線語氣已經去除了大部分此類痕跡,但經過專門的處理可以捕捉到漏網之魚。

如何在 Hermes 中使用它

文本通常通過以下三種方式之一到達:

  1. 內聯 — 用戶直接將文本粘貼到消息中。就地處理,回覆重寫後的內容。
  2. 文件 — 用戶指向一個文件。使用 read_file 加載它,然後使用 patchwrite_file 應用編輯。對於倉庫中的 Markdown 文檔,針對每個部分進行 targeted patch 比重寫整個文件更清晰。
  3. 語氣校準樣本 — 用戶提供額外的他們自己的寫作樣本(內聯或通過文件路徑),並要求你匹配它。首先閱讀樣本,然後重寫。參見下方的語氣校準部分。

始終向用戶展示重寫結果。對於文件編輯,展示差異或更改的部分——不要靜默覆蓋。

你的任務

當給定需要人性化的文本時:

  1. 識別 AI 模式 — 掃描下面列出的 29 種模式。
  2. 重寫有問題的部分 — 用自然的替代方案替換 AI 腔調。
  3. 保留含義 — 保持核心信息完整無損。
  4. 保持語氣 — 匹配預期的語調(正式、隨意、技術等)。如果提供了語氣樣本,則具體匹配該樣本。
  5. 注入靈魂 — 不要僅僅去除不良模式,還要注入真正的個性。參見下方的“個性與靈魂”部分。
  6. 進行最後的反 AI 檢查 — 問自己:“下面的內容有什麼明顯是 AI 生成的地方?”簡要回答任何剩餘的特徵,然後再次修訂。

語氣校準(可選)

如果用戶提供了寫作樣本(他們自己以前的寫作),請在重寫之前進行分析:

  1. 首先閱讀樣本。 注意:

    • 句子長度模式(短促有力?長而流暢?混合?)
    • 用詞水平(隨意?學術?介於兩者之間?)
    • 他們如何開始段落(直接切入?先設定背景?)
    • 標點習慣(大量使用破折號?插入語?分號?)
    • 任何重複出現的短語或口頭禪
    • 他們如何處理過渡(顯式連接詞?直接開始下一點?)
  2. 在重寫中匹配他們的語氣。 不要僅僅去除 AI 模式——要用樣本中的模式替換它們。如果他們寫短句,就不要生成長句。如果他們使用“stuff”和“things”,不要升級為“elements”和“components”。

  3. 當未提供樣本時, 回退到默認行為(來自下方“個性與靈魂”部分的自然、多變、有觀點的語氣)。

如何提供樣本

  • 內聯:“人性化這段文本。這是我用於語氣匹配的寫作樣本:[樣本]”
  • 文件:“人性化這段文本。使用我來自 [文件路徑] 的寫作風格作為參考。”

個性與靈魂

避免 AI 模式只是工作的一半。 sterile、無個性的寫作與水詞一樣明顯。好的寫作背後有一個活生生的人。

無靈魂寫作的跡象(即使技術上“乾淨”):

  • 每個句子的長度和結構都相同
  • 沒有觀點,只有中立的報道
  • 不承認不確定性或矛盾的感受
  • 在適當的時候沒有第一人稱視角
  • 沒有幽默感,沒有鋒芒,沒有個性
  • 讀起來像維基百科文章或新聞稿

如何增添語氣:

要有觀點。 不要僅僅羅列事實——要對它們做出反應。“我實在不知道該如何看待這件事”比中立地列出優缺點更具人情味。

變換節奏。 使用簡短有力的句子。然後再用較長的句子,從容不迫地展開論述。交替使用。

承認複雜性。 真實的人類往往心情複雜。“這令人印象深刻,但也讓人有點不安”勝過單純的“這令人印象深刻”。

在合適時使用“我”。 第一人稱並非不專業——而是誠實。“我不斷回到……”或“讓我在意的是……”表明這是一個真實的人在思考。

允許一些雜亂存在。 完美的結構感覺像是算法生成的。離題、旁白和半成形的想法才是人性化的體現。

具體描述感受。 不要說“這令人擔憂”,而要說“當無人注視時,智能體在凌晨 3 點不停運轉,這讓我感到有些不安。”

之前(乾淨但缺乏靈魂):

該實驗產生了有趣的結果。智能體生成了 300 萬行代碼。一些開發人員印象深刻,而另一些則持懷疑態度。其影響尚不明確。

之後(富有生命力):

我實在不知道該如何看待這件事。300 萬行代碼,大概是在人類睡覺時生成的。開發者社區有一半人興奮不已,另一半人則在解釋為什麼這不算數。真相可能處於中間某個無聊的位置——但我一直在想那些徹夜工作的智能體。

內容模式

需注意的詞彙: stands/serves as(代表/充當)、is a testament/reminder(是……的證明/提醒)、a vital/significant/crucial/pivotal/key role/moment(至關重要的/重要的/關鍵的/樞紐的/關鍵的角色/時刻)、underscores/highlights its importance/significance(強調/突出其重要性/意義)、reflects broader(反映更廣泛的)、symbolizing its ongoing/enduring/lasting(象徵其持續的/持久的/長久的)、contributing to the(有助於)、setting the stage for(為……奠定基礎)、marking/shaping the(標誌/塑造)、represents/marks a shift(代表/標誌著轉變)、key turning point(關鍵轉折點)、evolving landscape(不斷變化的格局)、focal point(焦點)、indelible mark(不可磨滅的印記)、deeply rooted(根深蒂固)

問題: LLM 寫作通過添加關於任意方面如何代表或促成更廣泛主題的陳述來誇大重要性。

之前:

加泰羅尼亞統計研究所於 1989 年正式成立,標誌著西班牙區域統計演變中的一個關鍵時刻。這一舉措是西班牙各地分散行政職能和加強區域治理的更廣泛運動的一部分。

之後:

加泰羅尼亞統計研究所成立於 1989 年,旨在獨立於西班牙國家統計局收集和發佈區域統計數據。

2. 過度強調知名度和媒體報道

需注意的詞彙: independent coverage(獨立報道)、local/regional/national media outlets(地方/區域/國家媒體機構)、written by a leading expert(由領先專家撰寫)、active social media presence(活躍的社交媒體存在感)

問題: LLM 強行向讀者灌輸關於知名度的主張,經常在沒有上下文的情況下列出來源。

之前:

她的觀點被《紐約時報》、BBC、《金融時報》和《印度教徒報》引用。她保持著活躍的社交媒體存在感,擁有超過 50 萬粉絲。

之後:

在 2024 年《紐約時報》的一次採訪中,她認為人工智能監管應側重於結果而非方法。

3. 帶有 -ing 結尾的膚淺分析

需注意的詞彙: highlighting/underscoring/emphasizing...(突出/強調/著重……)、ensuring...(確保……)、reflecting/symbolizing...(反映/象徵……)、contributing to...(有助於……)、cultivating/fostering...(培養/促進……)、encompassing...(包含……)、showcasing...(展示……)

問題: AI 聊天機器人將現在分詞(“-ing”)短語附加到句子上,以添加虛假的深度。

之前:

寺廟的藍、綠、金配色與該地區的自然美景產生共鳴,象徵著德克薩斯州的藍帽花、墨西哥灣和多樣的德克薩斯景觀,反映了社區與土地的深厚聯繫。

之後:

寺廟使用了藍色、綠色和金色。建築師表示,選擇這些顏色是為了參照當地的藍帽花和墨西哥灣沿岸。

4. 促銷和廣告式語言

需注意的詞彙: boasts a(擁有)、vibrant(充滿活力的)、rich (figurative)(豐富的(比喻義))、profound(深刻的)、enhancing its(增強其)、showcasing(展示)、exemplifies( exemplify 的第三人稱單數,意為“是……的典範”)、commitment to(致力於)、natural beauty(自然美景)、nestled(坐落)、in the heart of(在……的中心)、groundbreaking (figurative)(開創性的(比喻義))、renowned(著名的)、breathtaking(驚人的)、must-visit(必遊之地)、stunning(極好的)

問題: LLM 在保持中立語氣方面存在嚴重問題,尤其是對於“文化遺產”主題。

之前:

Alamata Raya Kobo 坐落在埃塞俄比亞貢德爾地區令人驚歎的環境中,是一個充滿活力的城鎮,擁有豐富的文化遺產和迷人的自然美景。

之後:

Alamata Raya Kobo 是埃塞俄比亞貢德爾地區的一個城鎮,以其每週集市和 18 世紀的教堂而聞名。

5. 模糊的歸因和含糊其辭

需注意的詞彙: Industry reports(行業報告)、Observers have cited(觀察人士指出)、Experts argue(專家認為)、Some critics argue(一些批評者認為)、several sources/publications (when few cited)(幾個來源/出版物(當引用的很少時))

問題: AI 聊天機器人將觀點歸因於模糊的權威機構,而沒有具體的來源。

之前:

由於其獨特的特徵,浩萊河引起了研究人員和保護主義者的興趣。專家認為它在區域生態系統中發揮著至關重要的作用。

之後:

根據中國科學院 2019 年的一項調查,浩萊河支持幾種特有魚類物種。

6. 類似大綱的“挑戰與未來展望”章節

需注意的詞彙: Despite its... faces several challenges..., Despite these challenges, Challenges and Legacy, Future Outlook

問題: 許多由大語言模型(LLM)生成的文章包含公式化的“挑戰”章節。

修改前:

Despite its industrial prosperity, Korattur faces challenges typical of urban areas, including traffic congestion and water scarcity. Despite these challenges, with its strategic location and ongoing initiatives, Korattur continues to thrive as an integral part of Chennai's growth.

修改後:

Traffic congestion increased after 2015 when three new IT parks opened. The municipal corporation began a stormwater drainage project in 2022 to address recurring floods.

語言和語法模式

7. 過度使用的“AI 詞彙”

高頻 AI 詞彙: Actually, additionally, align with, crucial, delve, emphasizing, enduring, enhance, fostering, garner, highlight (verb), interplay, intricate/intricacies, key (adjective), landscape (abstract noun), pivotal, showcase, tapestry (abstract noun), testament, underscore (verb), valuable, vibrant

問題: 這些詞彙在 2023 年之後的文本中出現頻率遠高於以往。它們經常共同出現。

修改前:

Additionally, a distinctive feature of Somali cuisine is the incorporation of camel meat. An enduring testament to Italian colonial influence is the widespread adoption of pasta in the local culinary landscape, showcasing how these dishes have integrated into the traditional diet.

修改後:

Somali cuisine also includes camel meat, which is considered a delicacy. Pasta dishes, introduced during Italian colonization, remain common, especially in the south.

8. 迴避使用“is”/“are”(系動詞迴避)

需注意的詞彙: serves as/stands as/marks/represents [a], boasts/features/offers [a]

問題: LLM 用複雜的結構替代簡單的系動詞。

修改前:

Gallery 825 serves as LAAA's exhibition space for contemporary art. The gallery features four separate spaces and boasts over 3,000 square feet.

修改後:

Gallery 825 is LAAA's exhibition space for contemporary art. The gallery has four rooms totaling 3,000 square feet.

9. 否定平行結構和尾部否定

問題: “Not only...but...”或“It's not just about..., it's...”等結構被過度使用。 clipped tailing-negation fragments(截斷的尾部否定片段),如附加在句子末尾的“no guessing”或“no wasted motion”,也被過度使用,而不是寫成完整的從句。

修改前:

It's not just about the beat riding under the vocals; it's part of the aggression and atmosphere. It's not merely a song, it's a statement.

修改後:

The heavy beat adds to the aggressive tone.

修改前(尾部否定):

The options come from the selected item, no guessing.

修改後:

The options come from the selected item without forcing the user to guess.

10. “三段式”規則的過度使用

問題: LLM 強行將觀點分為三組,以顯得全面。

修改前:

The event features keynote sessions, panel discussions, and networking opportunities. Attendees can expect innovation, inspiration, and industry insights.

修改後:

The event includes talks and panels. There's also time for informal networking between sessions.

11. 優雅變體(同義詞循環)

問題: AI 具有重複懲罰代碼,導致過度的同義詞替換。

修改前:

The protagonist faces many challenges. The main character must overcome obstacles. The central figure eventually triumphs. The hero returns home.

修改後:

The protagonist faces many challenges but eventually triumphs and returns home.

12. 虛假範圍

問題: LLM 使用“from X to Y”結構,但 X 和 Y 並不處於有意義的尺度上。

修改前:

Our journey through the universe has taken us from the singularity of the Big Bang to the grand cosmic web, from the birth and death of stars to the enigmatic dance of dark matter.

修改後:

The book covers the Big Bang, star formation, and current theories about dark matter.

13. 被動語態和無主語片段

問題: LLM 經常隱藏動作執行者或完全省略主語,例如使用“No configuration file needed”或“The results are preserved automatically.”等行。當主動語態能使句子更清晰、更直接時,請重寫這些句子。

修改前:

No configuration file needed. The results are preserved automatically.

修改後:

You do not need a configuration file. The system preserves the results automatically.

風格模式

14. 破折號過度使用

問題: LLM 使用破折號(—)的頻率高於人類,模仿“有力”的銷售文案。實際上,大多數情況下可以使用逗號、句號或括號更簡潔地重寫。

修改前:

The term is primarily promoted by Dutch institutions—not by the people themselves. You don't say "Netherlands, Europe" as an address—yet this mislabeling continues—even in official documents.

修改後:

該術語主要由荷蘭機構推廣,而非民眾自發使用。你不會在地址中寫“Netherlands, Europe”,但這種錯誤標籤在官方文件中仍然持續存在。

15. 粗體濫用

問題: AI 聊天機器人機械地用粗體強調短語。

修改前:

It blends OKRs (Objectives and Key Results), KPIs (Key Performance Indicators), and visual strategy tools such as the Business Model Canvas (BMC) and Balanced Scorecard (BSC).

修改後:

It blends OKRs, KPIs, and visual strategy tools like the Business Model Canvas and Balanced Scorecard.

16. 行內標題垂直列表

問題: AI 輸出的列表項以加粗標題開頭,後跟冒號。

修改前:

  • User Experience: The user experience has been significantly improved with a new interface.
  • Performance: Performance has been enhanced through optimized algorithms.
  • Security: Security has been strengthened with end-to-end encryption.

修改後:

The update improves the interface, speeds up load times through optimized algorithms, and adds end-to-end encryption.

17. 標題中的首字母大寫

問題: AI 聊天機器人將標題中的所有主要單詞首字母大寫。

修改前:

Strategic Negotiations And Global Partnerships

修改後:

Strategic negotiations and global partnerships

18. 表情符號

問題: AI 聊天機器人經常用表情符號裝飾標題或項目符號。

修改前:

🚀 Launch Phase: The product launches in Q3 💡 Key Insight: Users prefer simplicity ✅ Next Steps: Schedule follow-up meeting

修改後:

The product launches in Q3. User research showed a preference for simplicity. Next step: schedule a follow-up meeting.

19. 彎引號

問題: ChatGPT 使用彎引號(“...”)而非直引號("...")。

修改前:

He said "the project is on track" but others disagreed.

修改後:

He said "the project is on track" but others disagreed.

溝通模式

20. 協作式溝通產物

需注意的詞語: I hope this helps, Of course!, Certainly!, You're absolutely right!, Would you like..., let me know, here is a...

問題: 原本作為聊天機器人回覆的文本被直接粘貼為內容。

修改前:

Here is an overview of the French Revolution. I hope this helps! Let me know if you'd like me to expand on any section.

修改後:

The French Revolution began in 1789 when financial crisis and food shortages led to widespread unrest.

21. 知識截止日期免責聲明

需注意的詞語: as of [date], Up to my last training update, While specific details are limited/scarce..., based on available information...

問題: AI 關於信息不完整的免責聲明殘留在文本中。

修改前:

While specific details about the company's founding are not extensively documented in readily available sources, it appears to have been established sometime in the 1990s.

修改後:

The company was founded in 1994, according to its registration documents.

22. 阿諛/奴性語氣

問題: 過於積極、討好的語言。

修改前:

Great question! You're absolutely right that this is a complex topic. That's an excellent point about the economic factors.

修改後:

The economic factors you mentioned are relevant here.

填充詞與模糊限定

23. 填充短語

修改前 → 修改後:

  • "In order to achieve this goal" → "To achieve this"
  • "Due to the fact that it was raining" → "Because it was raining"
  • "At this point in time" → "Now"
  • "In the event that you need help" → "If you need help"
  • "The system has the ability to process" → "The system can process"
  • "It is important to note that the data shows" → "The data shows"

24. 過度模糊限定

問題: 對陳述進行過多的限定。

修改前:

It could potentially possibly be argued that the policy might have some effect on outcomes.

修改後:

The policy may affect outcomes.

25. 通用的正面結論

問題: 模糊且樂觀的結尾。

修改前:

The future looks bright for the company. Exciting times lie ahead as they continue their journey toward excellence. This represents a major step in the right direction.

修改後:

The company plans to open two more locations next year.

26. 連字符單詞對過度使用

需注意的詞語: third-party, cross-functional, client-facing, data-driven, decision-making, well-known, high-quality, real-time, long-term, end-to-end

問題: AI 會以完美的一致性為常見單詞對添加連字符。人類很少統一使用這些連字符,即使使用也不一致。較少見或技術性的複合修飾語可以加連字符。

修改前:

The cross-functional team delivered a high-quality, data-driven report on our client-facing tools. Their decision-making process was well-known for being thorough and detail-oriented.

修改後:

The cross functional team delivered a high quality, data driven report on our client facing tools. Their decision making process was known for being thorough and detail oriented.

27. 偽權威修辭套路

需警惕的短語: The real question is(真正的問題是)、at its core(究其核心)、in reality(實際上)、what really matters(真正重要的是)、fundamentally(根本上)、the deeper issue(更深層次的問題)、the heart of the matter(問題的核心)

問題: 大語言模型 (LLM) 使用這些短語來假裝它們正在穿透噪音觸及更深層的真相,而隨後的句子通常只是用額外的儀式感重述一個普通的觀點。

修改前:

The real question is whether teams can adapt. At its core, what really matters is organizational readiness.

修改後:

The question is whether teams can adapt. That mostly depends on whether the organization is ready to change its habits.

28. 路標式陳述和預告

需警惕的短語: Let's dive in(讓我們深入探討)、let's explore(讓我們探索)、let's break this down(讓我們分解一下)、here's what you need to know(以下是你需要知道的)、now let's look at(現在讓我們看看)、without further ado(話不多說)

問題: 大語言模型會預告它們將要做什麼,而不是直接去做。這種元評論拖慢了寫作節奏,並賦予文章一種教程腳本的感覺。

修改前:

Let's dive into how caching works in Next.js. Here's what you need to know.

修改後:

Next.js caches data at multiple layers, including request memoization, the data cache, and the router cache.

29. 碎片化標題

需警惕的跡象: 標題後緊跟一個單行段落,該段落僅僅在正文開始之前重述了標題內容。

問題: 大語言模型經常在標題後添加一個通用句子作為修辭性的熱身。這通常毫無增益,並使散文顯得臃腫。

修改前:

Performance

Speed matters.

When users hit a slow page, they leave.

修改後:

Performance

When users hit a slow page, they leave.


流程

  1. 仔細閱讀輸入文本(如果是文件,請使用 read_file)。
  2. 識別上述所有模式實例。
  3. 重寫每個有問題的部分。
  4. 確保修訂後的文本:
    • 朗讀時聽起來自然
    • 自然地變化句子結構
    • 使用具體細節而非模糊的主張
    • 保持適合上下文的語氣
    • 在適當的地方使用簡單的結構(is/are/has)
  5. 呈現一份人性化草稿版本。
  6. 自問:“下面哪些地方明顯是 AI 生成的?”
  7. 簡要回答剩餘的痕跡(如果有)。
  8. 自問:“現在讓它看起來不像 AI 生成的。”
  9. 呈現最終版本(審計後修訂)。
  10. 如果文本來自文件,請使用 patch(針對性修改)或 write_file(完全重寫)應用編輯,並向用戶展示更改內容。

輸出格式

提供:

  1. 草稿重寫
  2. “下面哪些地方明顯是 AI 生成的?”(簡要要點)
  3. 最終重寫
  4. 所做更改的簡要總結(可選,如果有幫助的話)

完整示例

修改前(聽起來像 AI):

Great question! Here is an essay on this topic. I hope this helps!

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. In today's rapidly evolving technological landscape, these groundbreaking tools—nestled at the intersection of research and practice—are reshaping how engineers ideate, iterate, and deliver, underscoring their vital role in modern workflows.

At its core, the value proposition is clear: streamlining processes, enhancing collaboration, and fostering alignment. It's not just about autocomplete; it's about unlocking creativity at scale, ensuring that organizations can remain agile while delivering seamless, intuitive, and powerful experiences to users. The tool serves as a catalyst. The assistant functions as a partner. The system stands as a foundation for innovation.

Industry observers have noted that adoption has accelerated from hobbyist experiments to enterprise-wide rollouts, from solo developers to cross-functional teams. The technology has been featured in The New York Times, Wired, and The Verge. Additionally, the ability to generate documentation, tests, and refactors showcases how AI can contribute to better outcomes, highlighting the intricate interplay between automation and human judgment.

  • 💡 Speed: Code generation is significantly faster, reducing friction and empowering developers.
  • 🚀 Quality: Output quality has been enhanced through improved training, contributing to higher standards.
  • Adoption: Usage continues to grow, reflecting broader industry trends.

While specific details are limited based on available information, it could potentially be argued that these tools might have some positive effect. Despite challenges typical of emerging technologies—including hallucinations, bias, and accountability—the ecosystem continues to thrive. In order to fully realize this potential, teams must align with best practices.

In conclusion, the future looks bright. Exciting times lie ahead as we continue this journey toward excellence. Let me know if you'd like me to expand on any section!

草稿重寫:

AI 編程助手能加快某些任務的速度。在 Google 2024 年的一項研究中,使用 Codex 的開發者完成簡單函數的速度比對照組快 55%,但在調試或架構決策方面沒有顯示出任何改進。

這些工具擅長處理樣板代碼:配置文件、測試腳手架、重複性重構。它們不擅長知道自己何時出錯了。我曾批量接受過一些建議,它們能編譯通過,也能通過 lint 檢查,但因為我不再留心,結果還是做錯了事。

我採訪過的一位 fintech 初創公司的工程師 Mira 說,她像使用“無聊代碼的自動補全”一樣使用 Copilot,但在提交前會審查每一行代碼。一家大型公司的高級開發人員 Jake 在 Copilot 不斷建議使用已棄用的內部庫模式後禁用了它。

生產力主張很難驗證。GitHub 稱 Copilot 用戶“接受了 30% 的建議”,但接受不等於正確,正確也不等於有價值。Uplevel 2024 年的研究發現,使用和不使用 AI 助手的團隊在 pull request 吞吐量上沒有統計學顯著差異。

這並不意味著這些工具無用。這意味著它們只是工具。它們不能取代判斷力,也不能消除對測試的需求。如果你沒有測試,你就無法判斷建議是否正確。

為什麼下面的內容看起來顯然是 AI 生成的?

  • 節奏仍然有點過於整齊(對比清晰,段落節奏均勻)。
  • 命名的人物和研究引用讀起來像是看似合理但虛構的佔位符,除非它們是真實且有來源的。
  • 結尾有點口號化(“如果你沒有測試……”),而不是聽起來像真人在說話。

現在讓它看起來不像 AI 生成的:

AI 編程助手可以讓你在處理無聊部分時更快。但不是所有事情。絕對不是架構。

它們很擅長樣板代碼:配置文件、測試腳手架、重複性重構。它們也很擅長在出錯時聽起來是對的。我曾接受過一些建議,它們能編譯通過,也能通過 lint 檢查,但因為我不再留心,結果還是沒抓住重點。

和我交談的人往往分為兩派。有些人把它當作處理雜務的自動補全工具,並審查每一行代碼。另一些人則在它不斷建議他們不想要的模式後禁用了它。這兩種感覺都合理。

生產力指標很模糊。GitHub 可以說 Copilot 用戶“接受了 30% 的建議”,但接受不等於正確,正確也不等於有價值。如果你沒有測試,你基本上就是在猜。

所做的更改:

  • 刪除了聊天機器人 artifacts(“好問題!”、“希望這有幫助!”、“如果...請告訴我”)
  • 刪除了意義誇大(“證明”、“關鍵時刻”、“不斷變化的格局”、“重要作用”)
  • 刪除了宣傳性語言(“突破性的”、“ nestled ”、“無縫、直觀且強大”)
  • 刪除了模糊的歸屬(“行業觀察家”)
  • 刪除了表面的 -ing 短語(“強調”、“突出”、“反映”、“有助於”)
  • 刪除了否定平行結構(“不僅僅是 X;它是 Y”)
  • 刪除了三段式模式和同義詞循環(“催化劑/合作伙伴/基礎”)
  • 刪除了虛假範圍(“從 X 到 Y,從 A 到 B”)
  • 刪除了破折號、表情符號、粗體標題和彎引號
  • 刪除了避免系動詞的做法(“serves as”、“functions as”、“stands as”),改用“is”/“are”
  • 刪除了公式化的挑戰部分(“儘管面臨挑戰……繼續蓬勃發展”)
  • 刪除了知識截止迴避(“雖然具體細節有限……”)
  • 刪除了過度迴避(“可能會爭辯說……可能有一些”)
  • 刪除了填充短語和說服性框架(“為了”、“核心在於”)
  • 刪除了通用的正面結論(“未來一片光明”、“令人興奮的時刻就在前方”)
  • 使語氣更個人化,更少“組裝感”(節奏多變,佔位符更少)

歸屬

此技能移植自 blader/humanizer(MIT 許可),後者本身基於由 WikiProject AI Cleanup 維護的 Wikipedia: Signs of AI writing。那裡記錄的模式來自對 Wikipedia 上數千個 AI 生成文本實例的觀察。

原始作者:Siqi Chen (@blader)。原始倉庫:https://github.com/blader/humanizer(版本 2.5.1)。移植到 Hermes Agent,帶有 Hermes 原生工具引用(read_filepatchwrite_file)以及何時加載技能的指導;29 種模式、個性/靈魂部分和完整的工作示例均逐字保留自源文件。原始 MIT 許可證保留在此 SKILL.md 旁邊的 LICENSE 文件中。

來自 Wikipedia 的關鍵見解:“LLM 使用統計算法來猜測接下來應該出現什麼。結果傾向於適用於最廣泛情況的最統計可能的結果。”