🔧 STEP 1. Prepare your materials 1. Organize your collected articles o Put all PDFs, DOCXs, or text files into one clearly labeled folder (e.g., AI_Summaries_Retention_Review/ ). o Rename each file with an informative label: Author_Year_TitleKey_Design.pdf → Example: Chen_2023_GPT4_Summary_24hRecall.pdf 2. Add brief notes or metadata o For each article, jot down in a separate sheet (Google Sheet or CSV): | ID | Author | Year | Design | Population | SummaryType | Comparator | Delay | Measures | Key Findings | | ---- | --------- | ------ | -------- | ------------- | -------------- | ------------ | --------- | ------------- | | 1 | Chen et al. | 2023 | Between | Undergrads | LLM Abstractive | Full Text | 24h | Free recall | Summaries ↓ Retention | This will help Notebook LM build better context and structured answers. STEP 2. Upload into Notebook LM 1. Open notebooklm.google.com → “ New Notebook ”. 2. Title it clearly: AI - Generated Summaries & Delayed Retention (College Students) 3. Upload your files : o Add all your articles as “sources.” o You can also paste your metadata sheet as a CSV or Google Doc link — Notebook LM will treat it as a data source. 💬 STEP 3. Set your “Context Preface” prompt Paste this into the “Notebook Overview” or “Context” section so Notebook LM always knows your focus: Context Preface: I am conducting a focused evidence synthesis on the question: Among college students, do AI - generated summaries weaken 24 - hour and 1 - week text retention compared to reading full - text materials? Please prioritize findings on delayed retention (≈ 24 h and 7 d) , summarize key moderators (summary type, parity of time - on - task, discipline, assessment type), and identify research gaps or inconsistencies. STEP 4. Run structured queries Ask Notebook LM in stages — here are the exact prompts to use. A. Descriptive mapping “Summarize for each uploaded article: participants, AI model used, summary type (abstractive/extractive), comparator, delay interval, outcome measure, and key findings on retention.” B. Evidence quality “Compare methodological quality across studies — note randomization, sample size, time - on - task control, and preregistration status.” C. Gap analysis “Based on all uploaded studies, list unresolved research gaps or inconsistencies regarding AI - generated summaries and delayed retention (24 h and 7 d). Group gaps by theme (design, population, mechanism, measurement).” D. Synthesis for writing “Write a 250 - word integrative summary of current evidence and the top 5 research gaps, in academic tone suitable for a literature review.” E. Study design inspiration “Propose 2 new experimental designs that would address the major gaps you identified, specifying independent/dependent variables, controls, and sample - size suggestions.” STEP 5. Export and refine • Use “ Export summary ” → Google Doc to start your literature review draft. • Copy key tables or graphs Notebook LM creates (e.g., a summary table of studies). • Manually verify all citations (Notebook LM sometimes mis - formats references).