繁中

TOPIC DETAIL

Complete AI Presentation Workflow

一套結合 AI 研究、萃取、設計、優化到展演思維的完整方法論,旨在產出兼具效率、視覺美感與人類獨特洞察的專業簡報。

核心觀念

  • AI 僅是 60 分的起點:AI 能快速生成結構完整、視覺尚可的簡報初稿,但若要達到專業水準,必須經由人類的「特殊洞察」與「個性表達」進行優化。
  • RED AI 流程:簡報製作分為三大步驟:Research (利用 Deep Research 搜索最新資料)、Extract (利用 NotebookLM 萃取核心觀點)、Design (利用 AI 工具生成視覺化內容)。
  • 認知負荷三原則:一份好的簡報應「降低內在負荷」(內容清晰易懂)、「去除外在負荷」(版面簡潔無干擾)、「增加本質負荷」(引發聽眾深度思考)。
  • 不可取代的價值:AI 無法複製的是你個人的專業、生活經驗、實作累積所形成的「特殊洞察」,以及展演時的知識、愉悅感與節奏所構成的「個性表達」(KH3R)。

整合工作流

  1. R (Research) - 深度研究
    • 在 Google Gemini 中使用「Deep Research」功能,針對簡報主題進行全面性、有來源根據的網路資料搜索。
    • 將生成的研究報告匯出至 Google 文件。
  2. E (Extract) - 萃取精華
    • 開啟 NotebookLM,建立新筆記本,將上述 Google 文件報告作為來源。
    • 透過向 NotebookLM 提問,逐步收斂,從報告中萃取出簡報所需的核心論點與「黃金筆記」。
  3. D (Design) - AI 初稿生成
    • 在 Google Gemini 中,使用「Canvas」功能,上傳純文字的簡報大綱(從黃金筆記整理而來)。
    • 給予明確提示詞(如「生成一份 Google Slides 簡報,風格要求專業、簡潔,使用藍色與灰色作為主色系」),讓 AI 生成「可編輯」的 Google 簡報初稿。
    • 另一條路:在 NotebookLM 中,定義好「視覺標準提示詞」(顏色、字體等),直接生成「圖片式」的簡報 PDF。
  4. 優化與修正
    • 文字校對:在 Google Slides 中,最優先、最重要的一步是逐頁確認所有文字內容的正確性。
    • 視覺美化:使用 Google Slide 內建的 Nano Banana Pro (Beautify) 功能,一鍵美化單頁投影片,將其轉換為高品質圖片。
    • 錯誤修正:對於 AI 生成圖片上的錯誤文字,利用 Nano Banana Pro 的「去除文字」功能抹除,再手動疊加正確的文字方塊。
  5. 動畫與展演
    • 將優化後的 Google 簡報下載為 PPTX,匯入 Canva。
    • 在 Canva 中添加更豐富的轉場動畫、元素動畫,提升視覺流暢度。
    • 結合「AI 簡報術最終回」的思維,在展演時注入個人魅力與特殊洞察,完成最後一哩路。

最佳金句

「真正危險不是被AI取代,而是我們表達的內容變得跟AI沒有分別。」 「AI只幫助你協助你的結構跟排版,而你要提供你的特殊洞察,在展演的時候,你必須要做好你的個性表達。」 「就算你用AI做的簡報,你也要修改到符合『降低內在負荷、去除外在負荷、增加本質負荷』這三個條件,這才是一個好的簡報。」

教學切入建議

此主題可作為一個完整的 AI 簡報工作坊。開頭先用 Gemini 快速生成一份 60 分的簡報,讓學員感受 AI 的速度。接著,引入 RED 流程,帶領學員一步步完成研究與萃取。然後,重點教學如何利用 Nano Banana Pro 和 Canva 將 60 分的初稿優化到 85 分。最後,以「AI 簡報術最終回」的思維作結,引導學員思考如何透過自身努力達到 95 分以上的水準。

常見誤區

  • 過度依賴 AI:拿到 AI 生成的初稿後,未經思考和修改就直接使用,導致簡報缺乏深度與個人觀點。
  • 順序錯誤:在確認文字內容正確之前,就先花大量時間做視覺美化,結果文字一改,所有美化工作都要重來。
  • 忽略聽眾:簡報內容只專注於自己想講的,沒有考慮聽眾的認知負荷與吸收效果。
  • 技術與內容脫節:只追求花俏的 AI 工具與動畫效果,但簡報的核心內容空洞,言之無物。

---en---

Core Concepts

  • AI as a 60-Point Starting Line: AI can quickly generate presentable draft presentations with decent visuals, but achieving professional quality requires human “unique insights” and “personalized expression” for optimization.
  • RED AI Process: Presentation creation is divided into three major steps: Research (using Deep Research to search for the latest information), Extract (using NotebookLM to extract core ideas), and Design (using AI tools to generate visual content).
  • Three Principles of Cognitive Load: A good presentation should “reduce intrinsic load” (content is clear and easy to understand), “remove extraneous load” (layout is clean and distraction-free), and “increase germane load” (promotes deeper thinking from the audience).
  • Irreplaceable Value: What AI cannot replicate is your personal professional knowledge, life experiences, and accumulated practical experience forming “unique insights,” as well as the “personalized expression” (KH3R) consisting of knowledge, happiness, and rhythm during the presentation.

Integrated Workflow

  1. R (Research) - Deep Research:
    • Use the “Deep Research” function in Google Gemini to conduct comprehensive, source-based online data searches for the presentation topic.
    • Export the generated research report to Google Docs.
  2. E (Extract) - Extract Essentials:
    • Open NotebookLM, create a new notebook, and use the Google Docs report as a source.
    • Through questioning NotebookLM, gradually refine and extract the core arguments and “golden notes” required for the presentation from the report.
  3. D (Design) - AI Draft Generation:
    • In Google Gemini, use the “Canvas” function and upload a plain text outline of the presentation (derived from the golden notes).
    • Provide clear prompts (e.g., “Generate a Google Slides presentation for me, with a professional and concise tech style, using blue and gray as primary colors”) to have AI generate an “editable” Google presentation draft.
    • Alternative Path: In NotebookLM, define “visual standard prompts” (colors, fonts, etc.) and directly generate a “picture-based” presentation PDF.
  4. Optimization and Correction:
    • Text Proofreading: In Google Slides, the highest priority and most important step is to review every page to ensure the accuracy of all text content.
    • Visual Enhancement: Use Google Slides’ built-in Nano Banana Pro (Beautify) function to instantly enhance a single slide, converting it into a high-quality image.
    • Error Correction: For incorrect text on AI-generated images, use Nano Banana Pro’s “remove text” function to erase it, then manually overlay correct text boxes.
  5. Animation and Presentation:
    • Download the optimized Google presentation as PPTX and import it into Canva.
    • Add richer transitions and element animations in Canva to enhance visual fluidity.
    • Incorporate the “AI Presentation Masterclass Finale” mindset, injecting personal charm and unique insights during the presentation to complete the final mile.

Best Quotes

“The real danger isn’t being replaced by AI, but our expressions becoming indistinguishable from AI’s.” “AI only assists you with structure and layout; you must provide your unique insights and embody your personalized expression during presentation.” “Even if your presentation is AI-generated, you must modify it to meet the three conditions of ‘reducing intrinsic load, removing extraneous load, and increasing germane load’ for it to be a good presentation.”

Teaching Entry Points

This topic can serve as a complete AI presentation workshop. Begin by having Gemini quickly generate a 60-point presentation, allowing students to experience AI’s speed. Then, introduce the RED process, guiding students step-by-step through research and extraction. Next, focus on teaching how to use Nano Banana Pro and Canva to optimize the 60-point draft to an 85-point level. Finally, conclude with the “AI Presentation Masterclass Finale” mindset, encouraging students to consider how to achieve a 95+ score through their own efforts.

Common Pitfalls

  • Over-reliance on AI: Using AI-generated drafts directly without critical thinking or modification, leading to presentations lacking depth and personal perspective.
  • Incorrect Order: Spending significant time on visual enhancements before confirming the accuracy of text content, resulting in rework when text changes are needed.
  • Ignoring the Audience: Focusing only on what one wants to say in a presentation, without considering the audience’s cognitive load and comprehension.
  • Disconnection Between Technology and Content: Solely pursuing flashy AI tools and animation effects, while the core content of the presentation remains empty and unsubstantial.

Key Concepts

  • AI 簡報
  • 特殊洞察
  • 個性表達
  • RED AI 流程
  • 認知負荷三原則
  • 不可取代的價值
  • Deep Research
  • NotebookLM
  • Google Gemini Canvas
  • Google Slides
  • Nano Banana Pro
  • Canva
  • PPTX
  • 人類靈魂
  • AI 效率

Related Episodes